# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=tahiti -run-pass=legalizer -global-isel-abort=0 %s -o - | FileCheck -check-prefix=SI  %s
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -run-pass=legalizer -global-isel-abort=0 %s -o - | FileCheck -check-prefix=VI %s
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -run-pass=legalizer -global-isel-abort=0 %s -o - | FileCheck -check-prefix=GFX9  %s
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -run-pass=legalizer -global-isel-abort=0 %s -o - | FileCheck -check-prefix=GFX9  %s

---
name: test_fsub_s32
body: |
  bb.0:
    liveins: $vgpr0, $vgpr1

    ; SI-LABEL: name: test_fsub_s32
    ; SI: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; SI: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
    ; SI: [[FSUB:%[0-9]+]]:_(s32) = G_FSUB [[COPY]], [[COPY1]]
    ; SI: $vgpr0 = COPY [[FSUB]](s32)
    ; VI-LABEL: name: test_fsub_s32
    ; VI: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; VI: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
    ; VI: [[FSUB:%[0-9]+]]:_(s32) = G_FSUB [[COPY]], [[COPY1]]
    ; VI: $vgpr0 = COPY [[FSUB]](s32)
    ; GFX9-LABEL: name: test_fsub_s32
    ; GFX9: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; GFX9: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
    ; GFX9: [[FSUB:%[0-9]+]]:_(s32) = G_FSUB [[COPY]], [[COPY1]]
    ; GFX9: $vgpr0 = COPY [[FSUB]](s32)
    %0:_(s32) = COPY $vgpr0
    %1:_(s32) = COPY $vgpr1
    %2:_(s32) = G_FSUB %0, %1
    $vgpr0 = COPY %2
...
---
name: test_fsub_s64
body: |
  bb.0:
    liveins: $vgpr0_vgpr1, $vgpr2_vgpr3

    ; SI-LABEL: name: test_fsub_s64
    ; SI: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
    ; SI: [[COPY1:%[0-9]+]]:_(s64) = COPY $vgpr2_vgpr3
    ; SI: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[COPY1]]
    ; SI: [[FADD:%[0-9]+]]:_(s64) = G_FADD [[COPY]], [[FNEG]]
    ; SI: $vgpr0_vgpr1 = COPY [[FADD]](s64)
    ; VI-LABEL: name: test_fsub_s64
    ; VI: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
    ; VI: [[COPY1:%[0-9]+]]:_(s64) = COPY $vgpr2_vgpr3
    ; VI: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[COPY1]]
    ; VI: [[FADD:%[0-9]+]]:_(s64) = G_FADD [[COPY]], [[FNEG]]
    ; VI: $vgpr0_vgpr1 = COPY [[FADD]](s64)
    ; GFX9-LABEL: name: test_fsub_s64
    ; GFX9: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
    ; GFX9: [[COPY1:%[0-9]+]]:_(s64) = COPY $vgpr2_vgpr3
    ; GFX9: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[COPY1]]
    ; GFX9: [[FADD:%[0-9]+]]:_(s64) = G_FADD [[COPY]], [[FNEG]]
    ; GFX9: $vgpr0_vgpr1 = COPY [[FADD]](s64)
    %0:_(s64) = COPY $vgpr0_vgpr1
    %1:_(s64) = COPY $vgpr2_vgpr3
    %2:_(s64) = G_FSUB %0, %1
    $vgpr0_vgpr1 = COPY %2
...

---
name: test_fsub_s64_fmf
body: |
  bb.0:
    liveins: $vgpr0_vgpr1, $vgpr2_vgpr3

    ; SI-LABEL: name: test_fsub_s64_fmf
    ; SI: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
    ; SI: [[COPY1:%[0-9]+]]:_(s64) = COPY $vgpr2_vgpr3
    ; SI: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[COPY1]]
    ; SI: [[FADD:%[0-9]+]]:_(s64) = nnan nsz G_FADD [[COPY]], [[FNEG]]
    ; SI: $vgpr0_vgpr1 = COPY [[FADD]](s64)
    ; VI-LABEL: name: test_fsub_s64_fmf
    ; VI: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
    ; VI: [[COPY1:%[0-9]+]]:_(s64) = COPY $vgpr2_vgpr3
    ; VI: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[COPY1]]
    ; VI: [[FADD:%[0-9]+]]:_(s64) = nnan nsz G_FADD [[COPY]], [[FNEG]]
    ; VI: $vgpr0_vgpr1 = COPY [[FADD]](s64)
    ; GFX9-LABEL: name: test_fsub_s64_fmf
    ; GFX9: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
    ; GFX9: [[COPY1:%[0-9]+]]:_(s64) = COPY $vgpr2_vgpr3
    ; GFX9: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[COPY1]]
    ; GFX9: [[FADD:%[0-9]+]]:_(s64) = nnan nsz G_FADD [[COPY]], [[FNEG]]
    ; GFX9: $vgpr0_vgpr1 = COPY [[FADD]](s64)
    %0:_(s64) = COPY $vgpr0_vgpr1
    %1:_(s64) = COPY $vgpr2_vgpr3
    %2:_(s64) = nnan nsz G_FSUB %0, %1
    $vgpr0_vgpr1 = COPY %2
...

---
name: test_fsub_s16
body: |
  bb.0:
    liveins: $vgpr0, $vgpr1

    ; SI-LABEL: name: test_fsub_s16
    ; SI: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; SI: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
    ; SI: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s32)
    ; SI: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY1]](s32)
    ; SI: [[FNEG:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC1]]
    ; SI: [[FPEXT:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC]](s16)
    ; SI: [[FPEXT1:%[0-9]+]]:_(s32) = G_FPEXT [[FNEG]](s16)
    ; SI: [[FADD:%[0-9]+]]:_(s32) = G_FADD [[FPEXT]], [[FPEXT1]]
    ; SI: [[FPTRUNC:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD]](s32)
    ; SI: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FPTRUNC]](s16)
    ; SI: $vgpr0 = COPY [[ANYEXT]](s32)
    ; VI-LABEL: name: test_fsub_s16
    ; VI: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; VI: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
    ; VI: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s32)
    ; VI: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY1]](s32)
    ; VI: [[FNEG:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC1]]
    ; VI: [[FADD:%[0-9]+]]:_(s16) = G_FADD [[TRUNC]], [[FNEG]]
    ; VI: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FADD]](s16)
    ; VI: $vgpr0 = COPY [[ANYEXT]](s32)
    ; GFX9-LABEL: name: test_fsub_s16
    ; GFX9: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
    ; GFX9: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
    ; GFX9: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s32)
    ; GFX9: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY1]](s32)
    ; GFX9: [[FNEG:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC1]]
    ; GFX9: [[FADD:%[0-9]+]]:_(s16) = G_FADD [[TRUNC]], [[FNEG]]
    ; GFX9: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FADD]](s16)
    ; GFX9: $vgpr0 = COPY [[ANYEXT]](s32)
    %0:_(s32) = COPY $vgpr0
    %1:_(s32) = COPY $vgpr1
    %2:_(s16) = G_TRUNC %0
    %3:_(s16) = G_TRUNC %1

    %4:_(s16) = G_FSUB %2, %3
    %5:_(s32) = G_ANYEXT %4
    $vgpr0 = COPY %5
...

---
name: test_fsub_v2s32
body: |
  bb.0.entry:
    liveins: $vgpr0_vgpr1, $vgpr2_vgpr3

    ; SI-LABEL: name: test_fsub_v2s32
    ; SI: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; SI: [[COPY1:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
    ; SI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; SI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<2 x s32>)
    ; SI: [[FSUB:%[0-9]+]]:_(s32) = G_FSUB [[UV]], [[UV2]]
    ; SI: [[FSUB1:%[0-9]+]]:_(s32) = G_FSUB [[UV1]], [[UV3]]
    ; SI: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[FSUB]](s32), [[FSUB1]](s32)
    ; SI: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
    ; VI-LABEL: name: test_fsub_v2s32
    ; VI: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; VI: [[COPY1:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
    ; VI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; VI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<2 x s32>)
    ; VI: [[FSUB:%[0-9]+]]:_(s32) = G_FSUB [[UV]], [[UV2]]
    ; VI: [[FSUB1:%[0-9]+]]:_(s32) = G_FSUB [[UV1]], [[UV3]]
    ; VI: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[FSUB]](s32), [[FSUB1]](s32)
    ; VI: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
    ; GFX9-LABEL: name: test_fsub_v2s32
    ; GFX9: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; GFX9: [[COPY1:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
    ; GFX9: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; GFX9: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<2 x s32>)
    ; GFX9: [[FSUB:%[0-9]+]]:_(s32) = G_FSUB [[UV]], [[UV2]]
    ; GFX9: [[FSUB1:%[0-9]+]]:_(s32) = G_FSUB [[UV1]], [[UV3]]
    ; GFX9: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[FSUB]](s32), [[FSUB1]](s32)
    ; GFX9: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
    %0:_(<2 x s32>) = COPY $vgpr0_vgpr1
    %1:_(<2 x s32>) = COPY $vgpr2_vgpr3
    %2:_(<2 x s32>) = G_FSUB %0, %1
    $vgpr0_vgpr1 = COPY %2
...

---
name: test_fsub_v2s32_flags
body: |
  bb.0.entry:
    liveins: $vgpr0_vgpr1, $vgpr2_vgpr3

    ; SI-LABEL: name: test_fsub_v2s32_flags
    ; SI: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; SI: [[COPY1:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
    ; SI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; SI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<2 x s32>)
    ; SI: [[FSUB:%[0-9]+]]:_(s32) = nnan G_FSUB [[UV]], [[UV2]]
    ; SI: [[FSUB1:%[0-9]+]]:_(s32) = nnan G_FSUB [[UV1]], [[UV3]]
    ; SI: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[FSUB]](s32), [[FSUB1]](s32)
    ; SI: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
    ; VI-LABEL: name: test_fsub_v2s32_flags
    ; VI: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; VI: [[COPY1:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
    ; VI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; VI: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<2 x s32>)
    ; VI: [[FSUB:%[0-9]+]]:_(s32) = nnan G_FSUB [[UV]], [[UV2]]
    ; VI: [[FSUB1:%[0-9]+]]:_(s32) = nnan G_FSUB [[UV1]], [[UV3]]
    ; VI: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[FSUB]](s32), [[FSUB1]](s32)
    ; VI: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
    ; GFX9-LABEL: name: test_fsub_v2s32_flags
    ; GFX9: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
    ; GFX9: [[COPY1:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
    ; GFX9: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
    ; GFX9: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<2 x s32>)
    ; GFX9: [[FSUB:%[0-9]+]]:_(s32) = nnan G_FSUB [[UV]], [[UV2]]
    ; GFX9: [[FSUB1:%[0-9]+]]:_(s32) = nnan G_FSUB [[UV1]], [[UV3]]
    ; GFX9: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[FSUB]](s32), [[FSUB1]](s32)
    ; GFX9: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
    %0:_(<2 x s32>) = COPY $vgpr0_vgpr1
    %1:_(<2 x s32>) = COPY $vgpr2_vgpr3
    %2:_(<2 x s32>) = nnan G_FSUB %0, %1
    $vgpr0_vgpr1 = COPY %2
...

---
name: test_fsub_v3s32
body: |
  bb.0.entry:
    liveins: $vgpr0_vgpr1_vgpr2, $vgpr3_vgpr4_vgpr5

    ; SI-LABEL: name: test_fsub_v3s32
    ; SI: [[COPY:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
    ; SI: [[COPY1:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr3_vgpr4_vgpr5
    ; SI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<3 x s32>)
    ; SI: [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<3 x s32>)
    ; SI: [[FSUB:%[0-9]+]]:_(s32) = G_FSUB [[UV]], [[UV3]]
    ; SI: [[FSUB1:%[0-9]+]]:_(s32) = G_FSUB [[UV1]], [[UV4]]
    ; SI: [[FSUB2:%[0-9]+]]:_(s32) = G_FSUB [[UV2]], [[UV5]]
    ; SI: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[FSUB]](s32), [[FSUB1]](s32), [[FSUB2]](s32)
    ; SI: $vgpr0_vgpr1_vgpr2 = COPY [[BUILD_VECTOR]](<3 x s32>)
    ; VI-LABEL: name: test_fsub_v3s32
    ; VI: [[COPY:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
    ; VI: [[COPY1:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr3_vgpr4_vgpr5
    ; VI: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<3 x s32>)
    ; VI: [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<3 x s32>)
    ; VI: [[FSUB:%[0-9]+]]:_(s32) = G_FSUB [[UV]], [[UV3]]
    ; VI: [[FSUB1:%[0-9]+]]:_(s32) = G_FSUB [[UV1]], [[UV4]]
    ; VI: [[FSUB2:%[0-9]+]]:_(s32) = G_FSUB [[UV2]], [[UV5]]
    ; VI: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[FSUB]](s32), [[FSUB1]](s32), [[FSUB2]](s32)
    ; VI: $vgpr0_vgpr1_vgpr2 = COPY [[BUILD_VECTOR]](<3 x s32>)
    ; GFX9-LABEL: name: test_fsub_v3s32
    ; GFX9: [[COPY:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
    ; GFX9: [[COPY1:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr3_vgpr4_vgpr5
    ; GFX9: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<3 x s32>)
    ; GFX9: [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<3 x s32>)
    ; GFX9: [[FSUB:%[0-9]+]]:_(s32) = G_FSUB [[UV]], [[UV3]]
    ; GFX9: [[FSUB1:%[0-9]+]]:_(s32) = G_FSUB [[UV1]], [[UV4]]
    ; GFX9: [[FSUB2:%[0-9]+]]:_(s32) = G_FSUB [[UV2]], [[UV5]]
    ; GFX9: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[FSUB]](s32), [[FSUB1]](s32), [[FSUB2]](s32)
    ; GFX9: $vgpr0_vgpr1_vgpr2 = COPY [[BUILD_VECTOR]](<3 x s32>)
    %0:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
    %1:_(<3 x s32>) = COPY $vgpr3_vgpr4_vgpr5
    %2:_(<3 x s32>) = G_FSUB %0, %1
    $vgpr0_vgpr1_vgpr2 = COPY %2
...

---
name: test_fsub_v2s64
body: |
  bb.0.entry:
    liveins: $vgpr0_vgpr1_vgpr2_vgpr3, $vgpr4_vgpr5_vgpr6_vgpr7

    ; SI-LABEL: name: test_fsub_v2s64
    ; SI: [[COPY:%[0-9]+]]:_(<2 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
    ; SI: [[COPY1:%[0-9]+]]:_(<2 x s64>) = COPY $vgpr4_vgpr5_vgpr6_vgpr7
    ; SI: [[UV:%[0-9]+]]:_(s64), [[UV1:%[0-9]+]]:_(s64) = G_UNMERGE_VALUES [[COPY]](<2 x s64>)
    ; SI: [[UV2:%[0-9]+]]:_(s64), [[UV3:%[0-9]+]]:_(s64) = G_UNMERGE_VALUES [[COPY1]](<2 x s64>)
    ; SI: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[UV2]]
    ; SI: [[FADD:%[0-9]+]]:_(s64) = G_FADD [[UV]], [[FNEG]]
    ; SI: [[FNEG1:%[0-9]+]]:_(s64) = G_FNEG [[UV3]]
    ; SI: [[FADD1:%[0-9]+]]:_(s64) = G_FADD [[UV1]], [[FNEG1]]
    ; SI: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[FADD]](s64), [[FADD1]](s64)
    ; SI: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<2 x s64>)
    ; VI-LABEL: name: test_fsub_v2s64
    ; VI: [[COPY:%[0-9]+]]:_(<2 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
    ; VI: [[COPY1:%[0-9]+]]:_(<2 x s64>) = COPY $vgpr4_vgpr5_vgpr6_vgpr7
    ; VI: [[UV:%[0-9]+]]:_(s64), [[UV1:%[0-9]+]]:_(s64) = G_UNMERGE_VALUES [[COPY]](<2 x s64>)
    ; VI: [[UV2:%[0-9]+]]:_(s64), [[UV3:%[0-9]+]]:_(s64) = G_UNMERGE_VALUES [[COPY1]](<2 x s64>)
    ; VI: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[UV2]]
    ; VI: [[FADD:%[0-9]+]]:_(s64) = G_FADD [[UV]], [[FNEG]]
    ; VI: [[FNEG1:%[0-9]+]]:_(s64) = G_FNEG [[UV3]]
    ; VI: [[FADD1:%[0-9]+]]:_(s64) = G_FADD [[UV1]], [[FNEG1]]
    ; VI: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[FADD]](s64), [[FADD1]](s64)
    ; VI: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<2 x s64>)
    ; GFX9-LABEL: name: test_fsub_v2s64
    ; GFX9: [[COPY:%[0-9]+]]:_(<2 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
    ; GFX9: [[COPY1:%[0-9]+]]:_(<2 x s64>) = COPY $vgpr4_vgpr5_vgpr6_vgpr7
    ; GFX9: [[UV:%[0-9]+]]:_(s64), [[UV1:%[0-9]+]]:_(s64) = G_UNMERGE_VALUES [[COPY]](<2 x s64>)
    ; GFX9: [[UV2:%[0-9]+]]:_(s64), [[UV3:%[0-9]+]]:_(s64) = G_UNMERGE_VALUES [[COPY1]](<2 x s64>)
    ; GFX9: [[FNEG:%[0-9]+]]:_(s64) = G_FNEG [[UV2]]
    ; GFX9: [[FADD:%[0-9]+]]:_(s64) = G_FADD [[UV]], [[FNEG]]
    ; GFX9: [[FNEG1:%[0-9]+]]:_(s64) = G_FNEG [[UV3]]
    ; GFX9: [[FADD1:%[0-9]+]]:_(s64) = G_FADD [[UV1]], [[FNEG1]]
    ; GFX9: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s64>) = G_BUILD_VECTOR [[FADD]](s64), [[FADD1]](s64)
    ; GFX9: $vgpr0_vgpr1_vgpr2_vgpr3 = COPY [[BUILD_VECTOR]](<2 x s64>)
    %0:_(<2 x s64>) = COPY $vgpr0_vgpr1_vgpr2_vgpr3
    %1:_(<2 x s64>) = COPY $vgpr4_vgpr5_vgpr6_vgpr7
    %2:_(<2 x s64>) = G_FSUB %0, %1
    $vgpr0_vgpr1_vgpr2_vgpr3 = COPY %2
...

---
name: test_fsub_v2s16
body: |
  bb.0.entry:
    liveins: $vgpr0, $vgpr1

    ; SI-LABEL: name: test_fsub_v2s16
    ; SI: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
    ; SI: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
    ; SI: [[FNEG:%[0-9]+]]:_(<2 x s16>) = G_FNEG [[COPY1]]
    ; SI: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY]](<2 x s16>)
    ; SI: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; SI: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; SI: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; SI: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; SI: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[FNEG]](<2 x s16>)
    ; SI: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; SI: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; SI: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
    ; SI: [[FPEXT:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC]](s16)
    ; SI: [[FPEXT1:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC2]](s16)
    ; SI: [[FADD:%[0-9]+]]:_(s32) = G_FADD [[FPEXT]], [[FPEXT1]]
    ; SI: [[FPTRUNC:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD]](s32)
    ; SI: [[FPEXT2:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC1]](s16)
    ; SI: [[FPEXT3:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC3]](s16)
    ; SI: [[FADD1:%[0-9]+]]:_(s32) = G_FADD [[FPEXT2]], [[FPEXT3]]
    ; SI: [[FPTRUNC1:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD1]](s32)
    ; SI: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[FPTRUNC]](s16)
    ; SI: [[ZEXT1:%[0-9]+]]:_(s32) = G_ZEXT [[FPTRUNC1]](s16)
    ; SI: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[ZEXT1]], [[C]](s32)
    ; SI: [[OR:%[0-9]+]]:_(s32) = G_OR [[ZEXT]], [[SHL]]
    ; SI: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
    ; SI: $vgpr0 = COPY [[BITCAST2]](<2 x s16>)
    ; VI-LABEL: name: test_fsub_v2s16
    ; VI: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
    ; VI: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
    ; VI: [[FNEG:%[0-9]+]]:_(<2 x s16>) = G_FNEG [[COPY1]]
    ; VI: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[COPY]](<2 x s16>)
    ; VI: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; VI: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; VI: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; VI: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; VI: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[FNEG]](<2 x s16>)
    ; VI: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; VI: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; VI: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
    ; VI: [[FADD:%[0-9]+]]:_(s16) = G_FADD [[TRUNC]], [[TRUNC2]]
    ; VI: [[FADD1:%[0-9]+]]:_(s16) = G_FADD [[TRUNC1]], [[TRUNC3]]
    ; VI: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[FADD]](s16)
    ; VI: [[ZEXT1:%[0-9]+]]:_(s32) = G_ZEXT [[FADD1]](s16)
    ; VI: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[ZEXT1]], [[C]](s32)
    ; VI: [[OR:%[0-9]+]]:_(s32) = G_OR [[ZEXT]], [[SHL]]
    ; VI: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
    ; VI: $vgpr0 = COPY [[BITCAST2]](<2 x s16>)
    ; GFX9-LABEL: name: test_fsub_v2s16
    ; GFX9: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
    ; GFX9: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
    ; GFX9: [[FNEG:%[0-9]+]]:_(<2 x s16>) = G_FNEG [[COPY1]]
    ; GFX9: [[FADD:%[0-9]+]]:_(<2 x s16>) = G_FADD [[COPY]], [[FNEG]]
    ; GFX9: $vgpr0 = COPY [[FADD]](<2 x s16>)
    %0:_(<2 x s16>) = COPY $vgpr0
    %1:_(<2 x s16>) = COPY $vgpr1
    %2:_(<2 x s16>) = G_FSUB %0, %1
    $vgpr0 = COPY %2
...

---
name: test_fsub_v3s16
body: |
  bb.0:
    liveins: $vgpr0_vgpr1_vgpr2, $vgpr3_vgpr4_vgpr5
    ; CHECK-LABEL: name: test_or_v3s16
    ; CHECK: [[COPY:%[0-9]+]]:_(<6 x s16>) = COPY $vgpr0_vgpr1_vgpr2
    ; CHECK: [[COPY1:%[0-9]+]]:_(<6 x s16>) = COPY $vgpr3_vgpr4_vgpr5
    ; CHECK: [[UV:%[0-9]+]]:_(<3 x s16>), [[UV1:%[0-9]+]]:_(<3 x s16>) = G_UNMERGE_VALUES [[COPY]](<6 x s16>)
    ; CHECK: [[UV2:%[0-9]+]]:_(<3 x s16>), [[UV3:%[0-9]+]]:_(<3 x s16>) = G_UNMERGE_VALUES [[COPY1]](<6 x s16>)
    ; CHECK: [[DEF:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
    ; CHECK: [[INSERT:%[0-9]+]]:_(<4 x s16>) = G_INSERT [[DEF]], [[UV]](<3 x s16>), 0
    ; CHECK: [[INSERT1:%[0-9]+]]:_(<4 x s16>) = G_INSERT [[DEF]], [[UV2]](<3 x s16>), 0
    ; CHECK: [[OR:%[0-9]+]]:_(<4 x s16>) = G_OR [[INSERT]], [[INSERT1]]
    ; CHECK: [[EXTRACT:%[0-9]+]]:_(<3 x s16>) = G_EXTRACT [[OR]](<4 x s16>), 0
    ; CHECK: [[EXTRACT1:%[0-9]+]]:_(<3 x s16>) = G_EXTRACT [[DEF]](<4 x s16>), 0
    ; CHECK: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[EXTRACT]](<3 x s16>), [[EXTRACT1]](<3 x s16>)
    ; CHECK: $vgpr0_vgpr1_vgpr2 = COPY [[CONCAT_VECTORS]](<6 x s16>)
    ; SI-LABEL: name: test_fsub_v3s16
    ; SI: [[COPY:%[0-9]+]]:_(<6 x s16>) = COPY $vgpr0_vgpr1_vgpr2
    ; SI: [[COPY1:%[0-9]+]]:_(<6 x s16>) = COPY $vgpr3_vgpr4_vgpr5
    ; SI: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>), [[UV2:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY]](<6 x s16>)
    ; SI: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV]](<2 x s16>)
    ; SI: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; SI: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; SI: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; SI: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; SI: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>)
    ; SI: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; SI: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; SI: [[UV3:%[0-9]+]]:_(<2 x s16>), [[UV4:%[0-9]+]]:_(<2 x s16>), [[UV5:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY1]](<6 x s16>)
    ; SI: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[UV3]](<2 x s16>)
    ; SI: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST2]](s32)
    ; SI: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
    ; SI: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
    ; SI: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[UV4]](<2 x s16>)
    ; SI: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST3]](s32)
    ; SI: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C]](s32)
    ; SI: [[FNEG:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC3]]
    ; SI: [[FPEXT:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC]](s16)
    ; SI: [[FPEXT1:%[0-9]+]]:_(s32) = G_FPEXT [[FNEG]](s16)
    ; SI: [[FADD:%[0-9]+]]:_(s32) = G_FADD [[FPEXT]], [[FPEXT1]]
    ; SI: [[FPTRUNC:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD]](s32)
    ; SI: [[FNEG1:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC4]]
    ; SI: [[FPEXT2:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC1]](s16)
    ; SI: [[FPEXT3:%[0-9]+]]:_(s32) = G_FPEXT [[FNEG1]](s16)
    ; SI: [[FADD1:%[0-9]+]]:_(s32) = G_FADD [[FPEXT2]], [[FPEXT3]]
    ; SI: [[FPTRUNC1:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD1]](s32)
    ; SI: [[FNEG2:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC5]]
    ; SI: [[FPEXT4:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC2]](s16)
    ; SI: [[FPEXT5:%[0-9]+]]:_(s32) = G_FPEXT [[FNEG2]](s16)
    ; SI: [[FADD2:%[0-9]+]]:_(s32) = G_FADD [[FPEXT4]], [[FPEXT5]]
    ; SI: [[FPTRUNC2:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD2]](s32)
    ; SI: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
    ; SI: [[DEF1:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
    ; SI: [[DEF2:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
    ; SI: [[DEF3:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
    ; SI: [[UV6:%[0-9]+]]:_(<2 x s16>), [[UV7:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[DEF2]](<4 x s16>)
    ; SI: [[BITCAST4:%[0-9]+]]:_(s32) = G_BITCAST [[UV6]](<2 x s16>)
    ; SI: [[LSHR4:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST4]], [[C]](s32)
    ; SI: [[BITCAST5:%[0-9]+]]:_(s32) = G_BITCAST [[UV7]](<2 x s16>)
    ; SI: [[LSHR5:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST5]], [[C]](s32)
    ; SI: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[FPTRUNC]](s16)
    ; SI: [[ZEXT1:%[0-9]+]]:_(s32) = G_ZEXT [[FPTRUNC1]](s16)
    ; SI: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[ZEXT1]], [[C]](s32)
    ; SI: [[OR:%[0-9]+]]:_(s32) = G_OR [[ZEXT]], [[SHL]]
    ; SI: [[BITCAST6:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
    ; SI: [[ZEXT2:%[0-9]+]]:_(s32) = G_ZEXT [[FPTRUNC2]](s16)
    ; SI: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
    ; SI: [[COPY2:%[0-9]+]]:_(s32) = COPY [[BITCAST4]](s32)
    ; SI: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C1]]
    ; SI: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[AND]], [[C]](s32)
    ; SI: [[OR1:%[0-9]+]]:_(s32) = G_OR [[ZEXT2]], [[SHL1]]
    ; SI: [[BITCAST7:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
    ; SI: [[COPY3:%[0-9]+]]:_(s32) = COPY [[LSHR4]](s32)
    ; SI: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C1]]
    ; SI: [[COPY4:%[0-9]+]]:_(s32) = COPY [[BITCAST5]](s32)
    ; SI: [[AND2:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C1]]
    ; SI: [[SHL2:%[0-9]+]]:_(s32) = G_SHL [[AND2]], [[C]](s32)
    ; SI: [[OR2:%[0-9]+]]:_(s32) = G_OR [[AND1]], [[SHL2]]
    ; SI: [[BITCAST8:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR2]](s32)
    ; SI: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST6]](<2 x s16>), [[BITCAST7]](<2 x s16>), [[BITCAST8]](<2 x s16>)
    ; SI: $vgpr0_vgpr1_vgpr2 = COPY [[CONCAT_VECTORS]](<6 x s16>)
    ; VI-LABEL: name: test_fsub_v3s16
    ; VI: [[COPY:%[0-9]+]]:_(<6 x s16>) = COPY $vgpr0_vgpr1_vgpr2
    ; VI: [[COPY1:%[0-9]+]]:_(<6 x s16>) = COPY $vgpr3_vgpr4_vgpr5
    ; VI: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>), [[UV2:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY]](<6 x s16>)
    ; VI: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV]](<2 x s16>)
    ; VI: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; VI: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; VI: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; VI: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; VI: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>)
    ; VI: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; VI: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; VI: [[UV3:%[0-9]+]]:_(<2 x s16>), [[UV4:%[0-9]+]]:_(<2 x s16>), [[UV5:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY1]](<6 x s16>)
    ; VI: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[UV3]](<2 x s16>)
    ; VI: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST2]](s32)
    ; VI: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
    ; VI: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
    ; VI: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[UV4]](<2 x s16>)
    ; VI: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST3]](s32)
    ; VI: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C]](s32)
    ; VI: [[FNEG:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC3]]
    ; VI: [[FADD:%[0-9]+]]:_(s16) = G_FADD [[TRUNC]], [[FNEG]]
    ; VI: [[FNEG1:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC4]]
    ; VI: [[FADD1:%[0-9]+]]:_(s16) = G_FADD [[TRUNC1]], [[FNEG1]]
    ; VI: [[FNEG2:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC5]]
    ; VI: [[FADD2:%[0-9]+]]:_(s16) = G_FADD [[TRUNC2]], [[FNEG2]]
    ; VI: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
    ; VI: [[DEF1:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
    ; VI: [[DEF2:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
    ; VI: [[DEF3:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
    ; VI: [[UV6:%[0-9]+]]:_(<2 x s16>), [[UV7:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[DEF2]](<4 x s16>)
    ; VI: [[BITCAST4:%[0-9]+]]:_(s32) = G_BITCAST [[UV6]](<2 x s16>)
    ; VI: [[LSHR4:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST4]], [[C]](s32)
    ; VI: [[BITCAST5:%[0-9]+]]:_(s32) = G_BITCAST [[UV7]](<2 x s16>)
    ; VI: [[LSHR5:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST5]], [[C]](s32)
    ; VI: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[FADD]](s16)
    ; VI: [[ZEXT1:%[0-9]+]]:_(s32) = G_ZEXT [[FADD1]](s16)
    ; VI: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[ZEXT1]], [[C]](s32)
    ; VI: [[OR:%[0-9]+]]:_(s32) = G_OR [[ZEXT]], [[SHL]]
    ; VI: [[BITCAST6:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
    ; VI: [[ZEXT2:%[0-9]+]]:_(s32) = G_ZEXT [[FADD2]](s16)
    ; VI: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
    ; VI: [[COPY2:%[0-9]+]]:_(s32) = COPY [[BITCAST4]](s32)
    ; VI: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C1]]
    ; VI: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[AND]], [[C]](s32)
    ; VI: [[OR1:%[0-9]+]]:_(s32) = G_OR [[ZEXT2]], [[SHL1]]
    ; VI: [[BITCAST7:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
    ; VI: [[COPY3:%[0-9]+]]:_(s32) = COPY [[LSHR4]](s32)
    ; VI: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C1]]
    ; VI: [[COPY4:%[0-9]+]]:_(s32) = COPY [[BITCAST5]](s32)
    ; VI: [[AND2:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C1]]
    ; VI: [[SHL2:%[0-9]+]]:_(s32) = G_SHL [[AND2]], [[C]](s32)
    ; VI: [[OR2:%[0-9]+]]:_(s32) = G_OR [[AND1]], [[SHL2]]
    ; VI: [[BITCAST8:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR2]](s32)
    ; VI: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST6]](<2 x s16>), [[BITCAST7]](<2 x s16>), [[BITCAST8]](<2 x s16>)
    ; VI: $vgpr0_vgpr1_vgpr2 = COPY [[CONCAT_VECTORS]](<6 x s16>)
    ; GFX9-LABEL: name: test_fsub_v3s16
    ; GFX9: [[COPY:%[0-9]+]]:_(<6 x s16>) = COPY $vgpr0_vgpr1_vgpr2
    ; GFX9: [[COPY1:%[0-9]+]]:_(<6 x s16>) = COPY $vgpr3_vgpr4_vgpr5
    ; GFX9: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>), [[UV2:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY]](<6 x s16>)
    ; GFX9: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV]](<2 x s16>)
    ; GFX9: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; GFX9: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; GFX9: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; GFX9: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; GFX9: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>)
    ; GFX9: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; GFX9: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; GFX9: [[UV3:%[0-9]+]]:_(<2 x s16>), [[UV4:%[0-9]+]]:_(<2 x s16>), [[UV5:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY1]](<6 x s16>)
    ; GFX9: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[UV3]](<2 x s16>)
    ; GFX9: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST2]](s32)
    ; GFX9: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
    ; GFX9: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
    ; GFX9: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[UV4]](<2 x s16>)
    ; GFX9: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST3]](s32)
    ; GFX9: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C]](s32)
    ; GFX9: [[FNEG:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC3]]
    ; GFX9: [[FADD:%[0-9]+]]:_(s16) = G_FADD [[TRUNC]], [[FNEG]]
    ; GFX9: [[FNEG1:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC4]]
    ; GFX9: [[FADD1:%[0-9]+]]:_(s16) = G_FADD [[TRUNC1]], [[FNEG1]]
    ; GFX9: [[FNEG2:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC5]]
    ; GFX9: [[FADD2:%[0-9]+]]:_(s16) = G_FADD [[TRUNC2]], [[FNEG2]]
    ; GFX9: [[DEF:%[0-9]+]]:_(s16) = G_IMPLICIT_DEF
    ; GFX9: [[DEF1:%[0-9]+]]:_(<2 x s16>) = G_IMPLICIT_DEF
    ; GFX9: [[DEF2:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
    ; GFX9: [[DEF3:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
    ; GFX9: [[UV6:%[0-9]+]]:_(<2 x s16>), [[UV7:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[DEF2]](<4 x s16>)
    ; GFX9: [[BITCAST4:%[0-9]+]]:_(s32) = G_BITCAST [[UV6]](<2 x s16>)
    ; GFX9: [[LSHR4:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST4]], [[C]](s32)
    ; GFX9: [[BITCAST5:%[0-9]+]]:_(s32) = G_BITCAST [[UV7]](<2 x s16>)
    ; GFX9: [[LSHR5:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST5]], [[C]](s32)
    ; GFX9: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FADD]](s16)
    ; GFX9: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FADD1]](s16)
    ; GFX9: [[BUILD_VECTOR_TRUNC:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[ANYEXT]](s32), [[ANYEXT1]](s32)
    ; GFX9: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[FADD2]](s16)
    ; GFX9: [[COPY2:%[0-9]+]]:_(s32) = COPY [[BITCAST4]](s32)
    ; GFX9: [[BUILD_VECTOR_TRUNC1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[ANYEXT2]](s32), [[COPY2]](s32)
    ; GFX9: [[COPY3:%[0-9]+]]:_(s32) = COPY [[LSHR4]](s32)
    ; GFX9: [[COPY4:%[0-9]+]]:_(s32) = COPY [[BITCAST5]](s32)
    ; GFX9: [[BUILD_VECTOR_TRUNC2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY3]](s32), [[COPY4]](s32)
    ; GFX9: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR_TRUNC]](<2 x s16>), [[BUILD_VECTOR_TRUNC1]](<2 x s16>), [[BUILD_VECTOR_TRUNC2]](<2 x s16>)
    ; GFX9: $vgpr0_vgpr1_vgpr2 = COPY [[CONCAT_VECTORS]](<6 x s16>)
    %0:_(<6 x s16>) = COPY $vgpr0_vgpr1_vgpr2
    %1:_(<6 x s16>) = COPY $vgpr3_vgpr4_vgpr5
    %2:_(<3 x s16>), %3:_(<3 x s16>) = G_UNMERGE_VALUES %0
    %4:_(<3 x s16>), %5:_(<3 x s16>) = G_UNMERGE_VALUES %1
    %6:_(<3 x s16>) = G_FSUB %2, %4
    %7:_(<3 x s16>) = G_IMPLICIT_DEF
    %8:_(<6 x s16>) = G_CONCAT_VECTORS %6, %7
    $vgpr0_vgpr1_vgpr2 = COPY %8

...

---
name: test_fsub_v4s16
body: |
  bb.0.entry:
    liveins: $vgpr0_vgpr1, $vgpr2_vgpr3

    ; SI-LABEL: name: test_fsub_v4s16
    ; SI: [[COPY:%[0-9]+]]:_(<4 x s16>) = COPY $vgpr0_vgpr1
    ; SI: [[COPY1:%[0-9]+]]:_(<4 x s16>) = COPY $vgpr2_vgpr3
    ; SI: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY]](<4 x s16>)
    ; SI: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV]](<2 x s16>)
    ; SI: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; SI: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; SI: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; SI: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; SI: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>)
    ; SI: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; SI: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; SI: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
    ; SI: [[UV2:%[0-9]+]]:_(<2 x s16>), [[UV3:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY1]](<4 x s16>)
    ; SI: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[UV2]](<2 x s16>)
    ; SI: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST2]](s32)
    ; SI: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
    ; SI: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
    ; SI: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[UV3]](<2 x s16>)
    ; SI: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST3]](s32)
    ; SI: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C]](s32)
    ; SI: [[TRUNC7:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
    ; SI: [[FNEG:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC4]]
    ; SI: [[FPEXT:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC]](s16)
    ; SI: [[FPEXT1:%[0-9]+]]:_(s32) = G_FPEXT [[FNEG]](s16)
    ; SI: [[FADD:%[0-9]+]]:_(s32) = G_FADD [[FPEXT]], [[FPEXT1]]
    ; SI: [[FPTRUNC:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD]](s32)
    ; SI: [[FNEG1:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC5]]
    ; SI: [[FPEXT2:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC1]](s16)
    ; SI: [[FPEXT3:%[0-9]+]]:_(s32) = G_FPEXT [[FNEG1]](s16)
    ; SI: [[FADD1:%[0-9]+]]:_(s32) = G_FADD [[FPEXT2]], [[FPEXT3]]
    ; SI: [[FPTRUNC1:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD1]](s32)
    ; SI: [[FNEG2:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC6]]
    ; SI: [[FPEXT4:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC2]](s16)
    ; SI: [[FPEXT5:%[0-9]+]]:_(s32) = G_FPEXT [[FNEG2]](s16)
    ; SI: [[FADD2:%[0-9]+]]:_(s32) = G_FADD [[FPEXT4]], [[FPEXT5]]
    ; SI: [[FPTRUNC2:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD2]](s32)
    ; SI: [[FNEG3:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC7]]
    ; SI: [[FPEXT6:%[0-9]+]]:_(s32) = G_FPEXT [[TRUNC3]](s16)
    ; SI: [[FPEXT7:%[0-9]+]]:_(s32) = G_FPEXT [[FNEG3]](s16)
    ; SI: [[FADD3:%[0-9]+]]:_(s32) = G_FADD [[FPEXT6]], [[FPEXT7]]
    ; SI: [[FPTRUNC3:%[0-9]+]]:_(s16) = G_FPTRUNC [[FADD3]](s32)
    ; SI: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[FPTRUNC]](s16)
    ; SI: [[ZEXT1:%[0-9]+]]:_(s32) = G_ZEXT [[FPTRUNC1]](s16)
    ; SI: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[ZEXT1]], [[C]](s32)
    ; SI: [[OR:%[0-9]+]]:_(s32) = G_OR [[ZEXT]], [[SHL]]
    ; SI: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
    ; SI: [[ZEXT2:%[0-9]+]]:_(s32) = G_ZEXT [[FPTRUNC2]](s16)
    ; SI: [[ZEXT3:%[0-9]+]]:_(s32) = G_ZEXT [[FPTRUNC3]](s16)
    ; SI: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[ZEXT3]], [[C]](s32)
    ; SI: [[OR1:%[0-9]+]]:_(s32) = G_OR [[ZEXT2]], [[SHL1]]
    ; SI: [[BITCAST5:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
    ; SI: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST4]](<2 x s16>), [[BITCAST5]](<2 x s16>)
    ; SI: $vgpr0_vgpr1 = COPY [[CONCAT_VECTORS]](<4 x s16>)
    ; VI-LABEL: name: test_fsub_v4s16
    ; VI: [[COPY:%[0-9]+]]:_(<4 x s16>) = COPY $vgpr0_vgpr1
    ; VI: [[COPY1:%[0-9]+]]:_(<4 x s16>) = COPY $vgpr2_vgpr3
    ; VI: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY]](<4 x s16>)
    ; VI: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV]](<2 x s16>)
    ; VI: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; VI: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; VI: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; VI: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; VI: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>)
    ; VI: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; VI: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; VI: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
    ; VI: [[UV2:%[0-9]+]]:_(<2 x s16>), [[UV3:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY1]](<4 x s16>)
    ; VI: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[UV2]](<2 x s16>)
    ; VI: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST2]](s32)
    ; VI: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
    ; VI: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
    ; VI: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[UV3]](<2 x s16>)
    ; VI: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST3]](s32)
    ; VI: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C]](s32)
    ; VI: [[TRUNC7:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
    ; VI: [[FNEG:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC4]]
    ; VI: [[FADD:%[0-9]+]]:_(s16) = G_FADD [[TRUNC]], [[FNEG]]
    ; VI: [[FNEG1:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC5]]
    ; VI: [[FADD1:%[0-9]+]]:_(s16) = G_FADD [[TRUNC1]], [[FNEG1]]
    ; VI: [[FNEG2:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC6]]
    ; VI: [[FADD2:%[0-9]+]]:_(s16) = G_FADD [[TRUNC2]], [[FNEG2]]
    ; VI: [[FNEG3:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC7]]
    ; VI: [[FADD3:%[0-9]+]]:_(s16) = G_FADD [[TRUNC3]], [[FNEG3]]
    ; VI: [[ZEXT:%[0-9]+]]:_(s32) = G_ZEXT [[FADD]](s16)
    ; VI: [[ZEXT1:%[0-9]+]]:_(s32) = G_ZEXT [[FADD1]](s16)
    ; VI: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[ZEXT1]], [[C]](s32)
    ; VI: [[OR:%[0-9]+]]:_(s32) = G_OR [[ZEXT]], [[SHL]]
    ; VI: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
    ; VI: [[ZEXT2:%[0-9]+]]:_(s32) = G_ZEXT [[FADD2]](s16)
    ; VI: [[ZEXT3:%[0-9]+]]:_(s32) = G_ZEXT [[FADD3]](s16)
    ; VI: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[ZEXT3]], [[C]](s32)
    ; VI: [[OR1:%[0-9]+]]:_(s32) = G_OR [[ZEXT2]], [[SHL1]]
    ; VI: [[BITCAST5:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
    ; VI: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST4]](<2 x s16>), [[BITCAST5]](<2 x s16>)
    ; VI: $vgpr0_vgpr1 = COPY [[CONCAT_VECTORS]](<4 x s16>)
    ; GFX9-LABEL: name: test_fsub_v4s16
    ; GFX9: [[COPY:%[0-9]+]]:_(<4 x s16>) = COPY $vgpr0_vgpr1
    ; GFX9: [[COPY1:%[0-9]+]]:_(<4 x s16>) = COPY $vgpr2_vgpr3
    ; GFX9: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY]](<4 x s16>)
    ; GFX9: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV]](<2 x s16>)
    ; GFX9: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
    ; GFX9: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
    ; GFX9: [[LSHR:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST]], [[C]](s32)
    ; GFX9: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR]](s32)
    ; GFX9: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>)
    ; GFX9: [[TRUNC2:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
    ; GFX9: [[LSHR1:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST1]], [[C]](s32)
    ; GFX9: [[TRUNC3:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR1]](s32)
    ; GFX9: [[UV2:%[0-9]+]]:_(<2 x s16>), [[UV3:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[COPY1]](<4 x s16>)
    ; GFX9: [[BITCAST2:%[0-9]+]]:_(s32) = G_BITCAST [[UV2]](<2 x s16>)
    ; GFX9: [[TRUNC4:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST2]](s32)
    ; GFX9: [[LSHR2:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST2]], [[C]](s32)
    ; GFX9: [[TRUNC5:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR2]](s32)
    ; GFX9: [[BITCAST3:%[0-9]+]]:_(s32) = G_BITCAST [[UV3]](<2 x s16>)
    ; GFX9: [[TRUNC6:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST3]](s32)
    ; GFX9: [[LSHR3:%[0-9]+]]:_(s32) = G_LSHR [[BITCAST3]], [[C]](s32)
    ; GFX9: [[TRUNC7:%[0-9]+]]:_(s16) = G_TRUNC [[LSHR3]](s32)
    ; GFX9: [[FNEG:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC4]]
    ; GFX9: [[FADD:%[0-9]+]]:_(s16) = G_FADD [[TRUNC]], [[FNEG]]
    ; GFX9: [[FNEG1:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC5]]
    ; GFX9: [[FADD1:%[0-9]+]]:_(s16) = G_FADD [[TRUNC1]], [[FNEG1]]
    ; GFX9: [[FNEG2:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC6]]
    ; GFX9: [[FADD2:%[0-9]+]]:_(s16) = G_FADD [[TRUNC2]], [[FNEG2]]
    ; GFX9: [[FNEG3:%[0-9]+]]:_(s16) = G_FNEG [[TRUNC7]]
    ; GFX9: [[FADD3:%[0-9]+]]:_(s16) = G_FADD [[TRUNC3]], [[FNEG3]]
    ; GFX9: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[FADD]](s16)
    ; GFX9: [[ANYEXT1:%[0-9]+]]:_(s32) = G_ANYEXT [[FADD1]](s16)
    ; GFX9: [[BUILD_VECTOR_TRUNC:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[ANYEXT]](s32), [[ANYEXT1]](s32)
    ; GFX9: [[ANYEXT2:%[0-9]+]]:_(s32) = G_ANYEXT [[FADD2]](s16)
    ; GFX9: [[ANYEXT3:%[0-9]+]]:_(s32) = G_ANYEXT [[FADD3]](s16)
    ; GFX9: [[BUILD_VECTOR_TRUNC1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[ANYEXT2]](s32), [[ANYEXT3]](s32)
    ; GFX9: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR_TRUNC]](<2 x s16>), [[BUILD_VECTOR_TRUNC1]](<2 x s16>)
    ; GFX9: $vgpr0_vgpr1 = COPY [[CONCAT_VECTORS]](<4 x s16>)
    %0:_(<4 x s16>) = COPY $vgpr0_vgpr1
    %1:_(<4 x s16>) = COPY $vgpr2_vgpr3
    %2:_(<4 x s16>) = G_FSUB %0, %1
    $vgpr0_vgpr1 = COPY %2
...