llvm/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-ctpop.mir

# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -mtriple=amdgcn -mcpu=tahiti -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck %s

---
name: ctpop_s32_ss
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $sgpr0

    ; CHECK-LABEL: name: ctpop_s32_ss
    ; CHECK: liveins: $sgpr0
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
    ; CHECK-NEXT: [[S_BCNT1_I32_B32_:%[0-9]+]]:sreg_32 = S_BCNT1_I32_B32 [[COPY]], implicit-def dead $scc
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[S_BCNT1_I32_B32_]]
    %0:sgpr(s32) = COPY $sgpr0
    %1:sgpr(s32) = G_CTPOP %0
    S_ENDPGM 0, implicit %1
...

---
name: ctpop_s32_vs
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $sgpr0

    ; CHECK-LABEL: name: ctpop_s32_vs
    ; CHECK: liveins: $sgpr0
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
    ; CHECK-NEXT: [[V_BCNT_U32_B32_e64_:%[0-9]+]]:vgpr_32 = V_BCNT_U32_B32_e64 [[COPY]], 0, implicit $exec
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[V_BCNT_U32_B32_e64_]]
    %0:sgpr(s32) = COPY $sgpr0
    %1:vgpr(s32) = G_CTPOP %0
    S_ENDPGM 0, implicit %1
...

---
name: ctpop_s32_vv
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $vgpr0

    ; CHECK-LABEL: name: ctpop_s32_vv
    ; CHECK: liveins: $vgpr0
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
    ; CHECK-NEXT: [[V_BCNT_U32_B32_e64_:%[0-9]+]]:vgpr_32 = V_BCNT_U32_B32_e64 [[COPY]], 0, implicit $exec
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[V_BCNT_U32_B32_e64_]]
    %0:vgpr(s32) = COPY $vgpr0
    %1:vgpr(s32) = G_CTPOP %0
    S_ENDPGM 0, implicit %1
...

---
name: add_ctpop_s32_v_vv_commute0
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $vgpr0, $vgpr1

    ; CHECK-LABEL: name: add_ctpop_s32_v_vv_commute0
    ; CHECK: liveins: $vgpr0, $vgpr1
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
    ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
    ; CHECK-NEXT: [[V_BCNT_U32_B32_e64_:%[0-9]+]]:vgpr_32 = V_BCNT_U32_B32_e64 [[COPY]], 0, implicit $exec
    ; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64 = V_ADD_CO_U32_e64 [[V_BCNT_U32_B32_e64_]], [[COPY1]], 0, implicit $exec
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[V_ADD_CO_U32_e64_]]
    %0:vgpr(s32) = COPY $vgpr0
    %1:vgpr(s32) = COPY $vgpr1
    %2:vgpr(s32) = G_CTPOP %0
    %3:vgpr(s32) = G_ADD %2, %1
    S_ENDPGM 0, implicit %3
...

---
name: add_ctpop_s32_v_vv_commute1
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $vgpr0, $vgpr1

    ; CHECK-LABEL: name: add_ctpop_s32_v_vv_commute1
    ; CHECK: liveins: $vgpr0, $vgpr1
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
    ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr1
    ; CHECK-NEXT: [[V_BCNT_U32_B32_e64_:%[0-9]+]]:vgpr_32 = V_BCNT_U32_B32_e64 [[COPY]], 0, implicit $exec
    ; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64 = V_ADD_CO_U32_e64 [[COPY1]], [[V_BCNT_U32_B32_e64_]], 0, implicit $exec
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[V_ADD_CO_U32_e64_]]
    %0:vgpr(s32) = COPY $vgpr0
    %1:vgpr(s32) = COPY $vgpr1
    %2:vgpr(s32) = G_CTPOP %0
    %3:vgpr(s32) = G_ADD %1, %2
    S_ENDPGM 0, implicit %3
...

# Test add+ctpop pattern with all scalars. This should stay scalar.
---
name: add_ctpop_s32_s_ss_commute0
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $sgpr0, $sgpr1

    ; CHECK-LABEL: name: add_ctpop_s32_s_ss_commute0
    ; CHECK: liveins: $sgpr0, $sgpr1
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
    ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr1
    ; CHECK-NEXT: [[S_BCNT1_I32_B32_:%[0-9]+]]:sreg_32 = S_BCNT1_I32_B32 [[COPY]], implicit-def dead $scc
    ; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[S_BCNT1_I32_B32_]], [[COPY1]], implicit-def dead $scc
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[S_ADD_I32_]]
    %0:sgpr(s32) = COPY $sgpr0
    %1:sgpr(s32) = COPY $sgpr1
    %2:sgpr(s32) = G_CTPOP %0
    %3:sgpr(s32) = G_ADD %2, %1
    S_ENDPGM 0, implicit %3
...

---
name: add_ctpop_s32_v_vs_commute0
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $vgpr0, $sgpr0

    ; CHECK-LABEL: name: add_ctpop_s32_v_vs_commute0
    ; CHECK: liveins: $vgpr0, $sgpr0
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
    ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
    ; CHECK-NEXT: [[V_BCNT_U32_B32_e64_:%[0-9]+]]:vgpr_32 = V_BCNT_U32_B32_e64 [[COPY]], 0, implicit $exec
    ; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64 = V_ADD_CO_U32_e64 [[V_BCNT_U32_B32_e64_]], [[COPY1]], 0, implicit $exec
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[V_ADD_CO_U32_e64_]]
    %0:vgpr(s32) = COPY $vgpr0
    %1:sgpr(s32) = COPY $sgpr0
    %2:vgpr(s32) = G_CTPOP %0
    %3:vgpr(s32) = G_ADD %2, %1
    S_ENDPGM 0, implicit %3
...

# SGPR->VGPR ctpop with VALU add
---
name: add_ctpop_s32_v_sv_commute0
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $vgpr0, $sgpr0

    ; CHECK-LABEL: name: add_ctpop_s32_v_sv_commute0
    ; CHECK: liveins: $vgpr0, $sgpr0
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
    ; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr0
    ; CHECK-NEXT: [[V_BCNT_U32_B32_e64_:%[0-9]+]]:vgpr_32 = V_BCNT_U32_B32_e64 [[COPY1]], 0, implicit $exec
    ; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64 = V_ADD_CO_U32_e64 [[V_BCNT_U32_B32_e64_]], [[COPY]], 0, implicit $exec
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[V_ADD_CO_U32_e64_]]
    %0:vgpr(s32) = COPY $vgpr0
    %1:sgpr(s32) = COPY $sgpr0
    %2:vgpr(s32) = G_CTPOP %1
    %3:vgpr(s32) = G_ADD %2, %0
    S_ENDPGM 0, implicit %3
...

# Scalar ctpop with VALU add
---
name: add_ctpop_s32_s_sv_commute0
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $sgpr0, $vgpr0

    ; CHECK-LABEL: name: add_ctpop_s32_s_sv_commute0
    ; CHECK: liveins: $sgpr0, $vgpr0
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr0
    ; CHECK-NEXT: [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr0
    ; CHECK-NEXT: [[S_BCNT1_I32_B32_:%[0-9]+]]:sreg_32 = S_BCNT1_I32_B32 [[COPY]], implicit-def dead $scc
    ; CHECK-NEXT: [[V_ADD_CO_U32_e64_:%[0-9]+]]:vgpr_32, dead [[V_ADD_CO_U32_e64_1:%[0-9]+]]:sreg_64 = V_ADD_CO_U32_e64 [[S_BCNT1_I32_B32_]], [[COPY1]], 0, implicit $exec
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[V_ADD_CO_U32_e64_]]
    %0:sgpr(s32) = COPY $sgpr0
    %1:vgpr(s32) = COPY $vgpr0
    %2:sgpr(s32) = G_CTPOP %0
    %3:vgpr(s32) = G_ADD %2, %1
    S_ENDPGM 0, implicit %3
...

---
name: ctpop_s64_ss
legalized: true
regBankSelected: true
tracksRegLiveness: true

body: |
  bb.0:
    liveins: $sgpr0_sgpr1

    ; CHECK-LABEL: name: ctpop_s64_ss
    ; CHECK: liveins: $sgpr0_sgpr1
    ; CHECK-NEXT: {{  $}}
    ; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
    ; CHECK-NEXT: [[S_BCNT1_I32_B64_:%[0-9]+]]:sreg_32 = S_BCNT1_I32_B64 [[COPY]], implicit-def dead $scc
    ; CHECK-NEXT: S_ENDPGM 0, implicit [[S_BCNT1_I32_B64_]]
    %0:sgpr(s64) = COPY $sgpr0_sgpr1
    %1:sgpr(s32) = G_CTPOP %0
    S_ENDPGM 0, implicit %1
...