test/CodeGen/AMDGPU/merge-flat-with-global-load-store.mir

# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx90a -run-pass=si-load-store-opt -verify-machineinstrs -o - %s | FileCheck -check-prefix=GCN %s

---
name:            merge_flat_global_load_dword_2
body:             |
  bb.0.entry:

    ; GCN-LABEL: name: merge_flat_global_load_dword_2
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[FLAT_LOAD_DWORDX2_:%[0-9]+]]:vreg_64_align2 = FLAT_LOAD_DWORDX2 [[DEF]], 0, 0, implicit $exec, implicit $flat_scr :: (load (s64) from `ptr undef` + 4, align 4)
    ; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY [[FLAT_LOAD_DWORDX2_]].sub0
    ; GCN-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY killed [[FLAT_LOAD_DWORDX2_]].sub1
    ; GCN-NEXT: S_NOP 0, implicit [[COPY]], implicit [[COPY1]]
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = FLAT_LOAD_DWORD %0, 0, 0, implicit $exec, implicit $flat_scr :: (load (s32) from `float* undef` + 4, basealign 4)
    %2:vgpr_32 = GLOBAL_LOAD_DWORD %0, 4, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef` + 8, basealign 4, addrspace 1)
    S_NOP 0, implicit %1, implicit %2
...

---
name:            merge_global_flat_load_dword_2
body:             |
  bb.0.entry:

    ; GCN-LABEL: name: merge_global_flat_load_dword_2
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[FLAT_LOAD_DWORDX2_:%[0-9]+]]:vreg_64_align2 = FLAT_LOAD_DWORDX2 [[DEF]], 0, 0, implicit $exec, implicit $flat_scr :: (load (s64) from `ptr addrspace(1) undef`)
    ; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY [[FLAT_LOAD_DWORDX2_]].sub0
    ; GCN-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY killed [[FLAT_LOAD_DWORDX2_]].sub1
    ; GCN-NEXT: S_NOP 0, implicit [[COPY]], implicit [[COPY1]]
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = GLOBAL_LOAD_DWORD %0, 0, 0, implicit $exec :: (load (s32) from `float addrspace(1)* undef`, basealign 8, addrspace 1)
    %2:vgpr_32 = FLAT_LOAD_DWORD %0, 4, 0, implicit $exec, implicit $flat_scr :: (load (s32) from `i32* undef` + 4, basealign 8)
    S_NOP 0, implicit %1, implicit %2
...

---
name:            merge_global_flat_load_dword_3
body:             |
  bb.0.entry:

    ; GCN-LABEL: name: merge_global_flat_load_dword_3
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[FLAT_LOAD_DWORDX3_:%[0-9]+]]:vreg_96_align2 = FLAT_LOAD_DWORDX3 [[DEF]], 0, 0, implicit $exec, implicit $flat_scr :: (load (s96) from `ptr undef`, align 16)
    ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY [[FLAT_LOAD_DWORDX3_]].sub0_sub1
    ; GCN-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY killed [[FLAT_LOAD_DWORDX3_]].sub2
    ; GCN-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY [[COPY]].sub1
    ; GCN-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[COPY]].sub0
    ; GCN-NEXT: S_NOP 0, implicit [[COPY2]], implicit [[COPY3]]
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = GLOBAL_LOAD_DWORD %0, 4, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef` + 4, basealign 8, addrspace 1)
    %2:vgpr_32 = FLAT_LOAD_DWORD %0, 0, 0, implicit $exec, implicit $flat_scr :: (load (s32) from `float* undef`, basealign 16)
    %3:vgpr_32 = GLOBAL_LOAD_DWORD %0, 8, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef` + 8, basealign 8, addrspace 1)
    S_NOP 0, implicit %1, implicit %2
...

---
name:            merge_global_flat_load_dword_4
body:             |
  bb.0.entry:

    ; GCN-LABEL: name: merge_global_flat_load_dword_4
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[FLAT_LOAD_DWORDX4_:%[0-9]+]]:vreg_128_align2 = FLAT_LOAD_DWORDX4 [[DEF]], 4, 0, implicit $exec, implicit $flat_scr :: (load (s128) from `ptr addrspace(1) undef` + 4, align 4, basealign 8)
    ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_96_align2 = COPY [[FLAT_LOAD_DWORDX4_]].sub0_sub1_sub2
    ; GCN-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY killed [[FLAT_LOAD_DWORDX4_]].sub3
    ; GCN-NEXT: [[COPY2:%[0-9]+]]:vreg_64_align2 = COPY [[COPY]].sub0_sub1
    ; GCN-NEXT: [[COPY3:%[0-9]+]]:vgpr_32 = COPY killed [[COPY]].sub2
    ; GCN-NEXT: [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[COPY2]].sub0
    ; GCN-NEXT: [[COPY5:%[0-9]+]]:vgpr_32 = COPY killed [[COPY2]].sub1
    ; GCN-NEXT: S_NOP 0, implicit [[COPY4]], implicit [[COPY5]], implicit [[COPY3]], implicit [[COPY1]]
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = GLOBAL_LOAD_DWORD %0, 4, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef` + 4, basealign 8, addrspace 1)
    %2:vgpr_32 = GLOBAL_LOAD_DWORD %0, 8, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef` + 8, basealign 8, addrspace 1)
    %3:vgpr_32 = GLOBAL_LOAD_DWORD %0, 12, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef` + 12, basealign 8, addrspace 1)
    %4:vgpr_32 = FLAT_LOAD_DWORD %0, 16, 0, implicit $exec, implicit $flat_scr :: (load (s32) from `float* undef` + 16)
    S_NOP 0, implicit %1, implicit %2, implicit %3, implicit %4
...

---
name:            merge_flat_global_load_dwordx2
body:             |
  bb.0.entry:

    ; GCN-LABEL: name: merge_flat_global_load_dwordx2
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[FLAT_LOAD_DWORDX4_:%[0-9]+]]:vreg_128_align2 = FLAT_LOAD_DWORDX4 [[DEF]], 0, 0, implicit $exec, implicit $flat_scr :: (load (s128) from `ptr undef`, align 8)
    ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY [[FLAT_LOAD_DWORDX4_]].sub0_sub1
    ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY killed [[FLAT_LOAD_DWORDX4_]].sub2_sub3
    ; GCN-NEXT: S_NOP 0, implicit [[COPY]], implicit [[COPY1]]
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vreg_64_align2 = FLAT_LOAD_DWORDX2 %0, 0, 0, implicit $exec, implicit $flat_scr :: (load (s64) from `double* undef`)
    %2:vreg_64_align2 = GLOBAL_LOAD_DWORDX2 %0, 8, 0, implicit $exec :: (load (s64) from `i64 addrspace(1)* undef`, addrspace 1)
    S_NOP 0, implicit %1, implicit %2
...

---
name:            merge_flat_global_load_dwordx3
body:             |
  bb.0.entry:

    ; GCN-LABEL: name: merge_flat_global_load_dwordx3
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[FLAT_LOAD_DWORDX4_:%[0-9]+]]:vreg_128_align2 = FLAT_LOAD_DWORDX4 [[DEF]], 0, 0, implicit $exec, implicit $flat_scr :: (load (s128) from `ptr undef`, align 4)
    ; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY [[FLAT_LOAD_DWORDX4_]].sub0
    ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_96_align2 = COPY killed [[FLAT_LOAD_DWORDX4_]].sub1_sub2_sub3
    ; GCN-NEXT: S_NOP 0, implicit [[COPY]], implicit [[COPY1]]
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = FLAT_LOAD_DWORD %0, 0, 0, implicit $exec, implicit $flat_scr :: (load (s32) from `float* undef`)
    %2:vreg_96_align2 = GLOBAL_LOAD_DWORDX3 %0, 4, 0, implicit $exec :: (load (s96) from `<3 x i32> addrspace(1)* undef`, addrspace 1)
    S_NOP 0, implicit %1, implicit %2
...

---
name:            merge_global_flat_load_dwordx3
body:             |
  bb.0.entry:

    ; GCN-LABEL: name: merge_global_flat_load_dwordx3
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[FLAT_LOAD_DWORDX4_:%[0-9]+]]:vreg_128_align2 = FLAT_LOAD_DWORDX4 [[DEF]], 0, 0, implicit $exec, implicit $flat_scr :: (load (s128) from `ptr addrspace(1) undef`, align 4)
    ; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY [[FLAT_LOAD_DWORDX4_]].sub0
    ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_96_align2 = COPY killed [[FLAT_LOAD_DWORDX4_]].sub1_sub2_sub3
    ; GCN-NEXT: S_NOP 0, implicit [[COPY]], implicit [[COPY1]]
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = GLOBAL_LOAD_DWORD %0, 0, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef`, addrspace 1)
    %2:vreg_96_align2 = FLAT_LOAD_DWORDX3 %0, 4, 0, implicit $exec, implicit $flat_scr :: (load (s96) from `<3 x i32>* undef`)
    S_NOP 0, implicit %1, implicit %2
...

---
name:            no_merge_flat_global_load_dword_saddr
body:             |
  bb.0.entry:

    ; GCN-LABEL: name: no_merge_flat_global_load_dword_saddr
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
    ; GCN-NEXT: [[FLAT_LOAD_DWORD:%[0-9]+]]:vgpr_32 = FLAT_LOAD_DWORD [[DEF]], 0, 0, implicit $exec, implicit $flat_scr :: (load (s32) from `ptr undef`)
    ; GCN-NEXT: [[GLOBAL_LOAD_DWORDX2_SADDR:%[0-9]+]]:vreg_64_align2 = GLOBAL_LOAD_DWORDX2_SADDR [[DEF1]], [[DEF]].sub0, 4, 0, implicit $exec :: (load (s64) from `ptr addrspace(1) undef` + 4, align 4, addrspace 1)
    ; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY [[GLOBAL_LOAD_DWORDX2_SADDR]].sub0
    ; GCN-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY killed [[GLOBAL_LOAD_DWORDX2_SADDR]].sub1
    ; GCN-NEXT: S_NOP 0, implicit [[FLAT_LOAD_DWORD]], implicit [[COPY]], implicit [[COPY1]]
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:sreg_64_xexec = IMPLICIT_DEF
    %2:vgpr_32 = FLAT_LOAD_DWORD %0, 0, 0, implicit $exec, implicit $flat_scr :: (load (s32) from `float* undef`, basealign 4)
    %3:vgpr_32 = GLOBAL_LOAD_DWORD_SADDR %1, %0.sub0, 4, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef` + 4, basealign 4, addrspace 1)
    %4:vgpr_32 = GLOBAL_LOAD_DWORD_SADDR %1, %0.sub0, 8, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef` + 8, basealign 4, addrspace 1)
    S_NOP 0, implicit %2, implicit %3, implicit %4
...

---
name:            no_merge_global_saddr_flat_load_dword
body:             |
  bb.0.entry:

    ; GCN-LABEL: name: no_merge_global_saddr_flat_load_dword
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
    ; GCN-NEXT: [[GLOBAL_LOAD_DWORD_SADDR:%[0-9]+]]:vgpr_32 = GLOBAL_LOAD_DWORD_SADDR [[DEF1]], [[DEF]].sub0, 0, 0, implicit $exec :: (load (s32) from `ptr addrspace(1) undef`, addrspace 1)
    ; GCN-NEXT: [[FLAT_LOAD_DWORDX2_:%[0-9]+]]:vreg_64_align2 = FLAT_LOAD_DWORDX2 [[DEF]], 4, 0, implicit $exec, implicit $flat_scr :: (load (s64) from `ptr undef` + 4, align 4)
    ; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY [[FLAT_LOAD_DWORDX2_]].sub0
    ; GCN-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY killed [[FLAT_LOAD_DWORDX2_]].sub1
    ; GCN-NEXT: S_NOP 0, implicit [[GLOBAL_LOAD_DWORD_SADDR]], implicit [[COPY]], implicit [[COPY1]]
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:sreg_64_xexec = IMPLICIT_DEF
    %2:vgpr_32 = GLOBAL_LOAD_DWORD_SADDR %1, %0.sub0, 0, 0, implicit $exec :: (load (s32) from `i32 addrspace(1)* undef`, addrspace 1)
    %3:vgpr_32 = FLAT_LOAD_DWORD %0, 4, 0, implicit $exec, implicit $flat_scr :: (load (s32) from `i32* undef` + 4)
    %4:vgpr_32 = FLAT_LOAD_DWORD %0, 8, 0, implicit $exec, implicit $flat_scr :: (load (s32) from `i32* undef` + 8)
    S_NOP 0, implicit %2, implicit %3, implicit %4
...

---
name:            merge_flat_global_store_dword_2
body:             |
  bb.0.entry:
    ; GCN-LABEL: name: merge_flat_global_store_dword_2
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF2:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE killed [[DEF1]], %subreg.sub0, killed [[DEF2]], %subreg.sub1
    ; GCN-NEXT: FLAT_STORE_DWORDX2 [[DEF]], killed [[REG_SEQUENCE]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s64) into `ptr undef`, align 4)
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = IMPLICIT_DEF
    %2:vgpr_32 = IMPLICIT_DEF
    FLAT_STORE_DWORD %0, killed %1, 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into `i32* undef`)
    GLOBAL_STORE_DWORD killed %0, killed %2, 4, 0, implicit $exec :: (store (s32) into `i32 addrspace(1)* undef`, addrspace 1)
...

---
name:            merge_global_flat_store_dword_2
body:             |
  bb.0.entry:
    ; GCN-LABEL: name: merge_global_flat_store_dword_2
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF2:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64_align2 = REG_SEQUENCE killed [[DEF1]], %subreg.sub0, killed [[DEF2]], %subreg.sub1
    ; GCN-NEXT: FLAT_STORE_DWORDX2 [[DEF]], killed [[REG_SEQUENCE]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s64) into `ptr addrspace(1) undef`, align 4)
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = IMPLICIT_DEF
    %2:vgpr_32 = IMPLICIT_DEF
    GLOBAL_STORE_DWORD %0, killed %1, 0, 0, implicit $exec :: (store (s32) into `i32 addrspace(1)* undef`, addrspace 1)
    FLAT_STORE_DWORD %0, killed %2, 4, 0, implicit $exec, implicit $flat_scr :: (store (s32) into `i32* undef`)
...

---
name:            merge_flat_global_store_dwordx2
body:             |
  bb.0.entry:
    ; GCN-LABEL: name: merge_flat_global_store_dwordx2
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF2:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_96_align2 = REG_SEQUENCE [[DEF1]], %subreg.sub0, [[DEF2]], %subreg.sub1_sub2
    ; GCN-NEXT: FLAT_STORE_DWORDX3 [[DEF]], killed [[REG_SEQUENCE]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s96) into `ptr undef`, align 4)
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = IMPLICIT_DEF
    %2:vreg_64_align2 = IMPLICIT_DEF
    FLAT_STORE_DWORD %0, %1, 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into `i32* undef`)
    GLOBAL_STORE_DWORDX2 %0, %2, 4, 0, implicit $exec :: (store (s64) into `i64 addrspace(1)* undef`, addrspace 1)
...

---
name:            merge_flat_global_store_dwordx3
body:             |
  bb.0.entry:
    ; GCN-LABEL: name: merge_flat_global_store_dwordx3
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF2:%[0-9]+]]:vreg_96_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128_align2 = REG_SEQUENCE [[DEF1]], %subreg.sub0, [[DEF2]], %subreg.sub1_sub2_sub3
    ; GCN-NEXT: FLAT_STORE_DWORDX4 [[DEF]], killed [[REG_SEQUENCE]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s128) into `ptr undef`, align 4)
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = IMPLICIT_DEF
    %2:vreg_96_align2 = IMPLICIT_DEF
    FLAT_STORE_DWORD %0, %1, 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into `i32* undef`)
    GLOBAL_STORE_DWORDX3 %0, %2, 4, 0, implicit $exec :: (store (s96) into `<3 x i32> addrspace(1)* undef`, addrspace 1)
...

---
name:            merge_global_flat_store_dwordx2
body:             |
  bb.0.entry:
    ; GCN-LABEL: name: merge_global_flat_store_dwordx2
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF2:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_96_align2 = REG_SEQUENCE [[DEF1]], %subreg.sub2, [[DEF2]], %subreg.sub0_sub1
    ; GCN-NEXT: FLAT_STORE_DWORDX3 [[DEF]], killed [[REG_SEQUENCE]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s96) into `ptr undef`, align 8)
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = IMPLICIT_DEF
    %2:vreg_64_align2 = IMPLICIT_DEF
    GLOBAL_STORE_DWORD %0, %1, 8, 0, implicit $exec :: (store (s32) into `i32 addrspace(1)* undef`, addrspace 1)
    FLAT_STORE_DWORDX2 %0, %2, 0, 0, implicit $exec, implicit $flat_scr :: (store (s64) into `i64* undef`)
...

---
name:            merge_global_flat_store_dwordx3
body:             |
  bb.0.entry:
    ; GCN-LABEL: name: merge_global_flat_store_dwordx3
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF2:%[0-9]+]]:vreg_96_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_128_align2 = REG_SEQUENCE [[DEF1]], %subreg.sub3, [[DEF2]], %subreg.sub0_sub1_sub2
    ; GCN-NEXT: FLAT_STORE_DWORDX4 [[DEF]], killed [[REG_SEQUENCE]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s128) into `ptr undef`)
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:vgpr_32 = IMPLICIT_DEF
    %2:vreg_96_align2 = IMPLICIT_DEF
    GLOBAL_STORE_DWORD %0, %1, 12, 0, implicit $exec :: (store (s32) into `i32 addrspace(1)* undef`, addrspace 1)
    FLAT_STORE_DWORDX3 %0, %2, 0, 0, implicit $exec, implicit $flat_scr :: (store (s96) into `<3 x i32>* undef`)
...

---
name:            no_merge_flat_global_store_dword_saddr
body:             |
  bb.0.entry:
    ; GCN-LABEL: name: no_merge_flat_global_store_dword_saddr
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF2:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF3:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: FLAT_STORE_DWORD [[DEF]], [[DEF2]], 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into `ptr undef`)
    ; GCN-NEXT: GLOBAL_STORE_DWORD_SADDR [[DEF]].sub0, [[DEF3]], [[DEF1]], 4, 0, implicit $exec :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:sreg_64_xexec = IMPLICIT_DEF
    %2:vgpr_32 = IMPLICIT_DEF
    %3:vgpr_32 = IMPLICIT_DEF
    FLAT_STORE_DWORD %0, %2, 0, 0, implicit $exec, implicit $flat_scr :: (store (s32) into `i32* undef`)
    GLOBAL_STORE_DWORD_SADDR %0.sub0, %3, %1, 4, 0, implicit $exec :: (store (s32) into `i32 addrspace(1)* undef`, addrspace 1)
...

---
name:            no_merge_global_saddr_flat_store_dword
body:             |
  bb.0.entry:
    ; GCN-LABEL: name: no_merge_global_saddr_flat_store_dword
    ; GCN: [[DEF:%[0-9]+]]:vreg_64_align2 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF1:%[0-9]+]]:sreg_64_xexec = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF2:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: [[DEF3:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
    ; GCN-NEXT: GLOBAL_STORE_DWORD_SADDR [[DEF]].sub0, [[DEF2]], [[DEF1]], 0, 0, implicit $exec :: (store (s32) into `ptr addrspace(1) undef`, addrspace 1)
    ; GCN-NEXT: FLAT_STORE_DWORD [[DEF]], [[DEF3]], 4, 0, implicit $exec, implicit $flat_scr :: (store (s32) into `ptr undef`)
    %0:vreg_64_align2 = IMPLICIT_DEF
    %1:sreg_64_xexec = IMPLICIT_DEF
    %2:vgpr_32 = IMPLICIT_DEF
    %3:vgpr_32 = IMPLICIT_DEF
    GLOBAL_STORE_DWORD_SADDR %0.sub0, %2, %1, 0, 0, implicit $exec :: (store (s32) into `i32 addrspace(1)* undef`, addrspace 1)
    FLAT_STORE_DWORD %0, %3, 4, 0, implicit $exec, implicit $flat_scr :: (store (s32) into `i32* undef`)
...