test/CodeGen/AMDGPU/GlobalISel/inst-select-store-global.s96.mir

# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -march=amdgcn -mcpu=hawaii -run-pass=instruction-select -verify-machineinstrs  -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX7 %s
# RUN: llc -march=amdgcn -mcpu=hawaii -mattr=+flat-for-global -run-pass=instruction-select -verify-machineinstrs  -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX7-FLAT %s
# RUN: llc -march=amdgcn -mcpu=fiji -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX8 %s
# RUN: llc -march=amdgcn -mcpu=gfx900 -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX9 %s
# RUN: llc -march=amdgcn -mcpu=gfx1010 -run-pass=instruction-select -verify-machineinstrs -global-isel-abort=0 -o - %s | FileCheck -check-prefix=GFX10 %s

# 96-bit load/store is only legal for SI, so split the test out.

---

name: store_global_v3s32
legalized:       true
tracksRegLiveness: true
regBankSelected: true

body: |
  bb.0:
    liveins:  $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4

    ; GFX7-LABEL: name: store_global_v3s32
    ; GFX7: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX7: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
    ; GFX7: [[COPY1:%[0-9]+]]:vreg_96 = COPY $vgpr2_vgpr3_vgpr4
    ; GFX7: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 0
    ; GFX7: [[S_MOV_B32_1:%[0-9]+]]:sreg_32 = S_MOV_B32 61440
    ; GFX7: [[REG_SEQUENCE:%[0-9]+]]:sreg_64 = REG_SEQUENCE [[S_MOV_B32_]], %subreg.sub0, [[S_MOV_B32_1]], %subreg.sub1
    ; GFX7: [[S_MOV_B64_:%[0-9]+]]:sreg_64 = S_MOV_B64 0
    ; GFX7: [[REG_SEQUENCE1:%[0-9]+]]:sgpr_128 = REG_SEQUENCE [[S_MOV_B64_]], %subreg.sub0_sub1, [[REG_SEQUENCE]], %subreg.sub2_sub3
    ; GFX7: BUFFER_STORE_DWORDX3_ADDR64 [[COPY1]], [[COPY]], [[REG_SEQUENCE1]], 0, 0, 0, 0, 0, implicit $exec :: (store 12, align 16, addrspace 1)
    ; GFX7-FLAT-LABEL: name: store_global_v3s32
    ; GFX7-FLAT: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX7-FLAT: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
    ; GFX7-FLAT: [[COPY1:%[0-9]+]]:vreg_96 = COPY $vgpr2_vgpr3_vgpr4
    ; GFX7-FLAT: FLAT_STORE_DWORDX3 [[COPY]], [[COPY1]], 0, 0, implicit $exec, implicit $flat_scr :: (store 12, align 16, addrspace 1)
    ; GFX8-LABEL: name: store_global_v3s32
    ; GFX8: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX8: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
    ; GFX8: [[COPY1:%[0-9]+]]:vreg_96 = COPY $vgpr2_vgpr3_vgpr4
    ; GFX8: FLAT_STORE_DWORDX3 [[COPY]], [[COPY1]], 0, 0, implicit $exec, implicit $flat_scr :: (store 12, align 16, addrspace 1)
    ; GFX9-LABEL: name: store_global_v3s32
    ; GFX9: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX9: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
    ; GFX9: [[COPY1:%[0-9]+]]:vreg_96 = COPY $vgpr2_vgpr3_vgpr4
    ; GFX9: GLOBAL_STORE_DWORDX3 [[COPY]], [[COPY1]], 0, 0, implicit $exec :: (store 12, align 16, addrspace 1)
    ; GFX10-LABEL: name: store_global_v3s32
    ; GFX10: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX10: [[COPY:%[0-9]+]]:vreg_64 = COPY $vgpr0_vgpr1
    ; GFX10: [[COPY1:%[0-9]+]]:vreg_96 = COPY $vgpr2_vgpr3_vgpr4
    ; GFX10: GLOBAL_STORE_DWORDX3 [[COPY]], [[COPY1]], 0, 0, implicit $exec :: (store 12, align 16, addrspace 1)
    %0:vgpr(p1) = COPY $vgpr0_vgpr1
    %1:vgpr(<3 x s32>) = COPY $vgpr2_vgpr3_vgpr4
    G_STORE %1, %0 :: (store 12, align 16, addrspace 1)

...

---

name: store_global_s96
legalized:       true
tracksRegLiveness: true
regBankSelected: true

body: |
  bb.0:
    liveins:  $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4

    ; GFX6-LABEL: name: store_global_s96
    ; GFX6: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX6: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX6: [[COPY1:%[0-9]+]]:vgpr(s96) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX6: G_STORE [[COPY1]](s96), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX7-LABEL: name: store_global_s96
    ; GFX7: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX7: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX7: [[COPY1:%[0-9]+]]:vgpr(s96) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX7: G_STORE [[COPY1]](s96), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX7-FLAT-LABEL: name: store_global_s96
    ; GFX7-FLAT: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX7-FLAT: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX7-FLAT: [[COPY1:%[0-9]+]]:vgpr(s96) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX7-FLAT: G_STORE [[COPY1]](s96), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX8-LABEL: name: store_global_s96
    ; GFX8: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX8: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX8: [[COPY1:%[0-9]+]]:vgpr(s96) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX8: G_STORE [[COPY1]](s96), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX9-LABEL: name: store_global_s96
    ; GFX9: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX9: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX9: [[COPY1:%[0-9]+]]:vgpr(s96) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX9: G_STORE [[COPY1]](s96), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX10-LABEL: name: store_global_s96
    ; GFX10: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX10: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX10: [[COPY1:%[0-9]+]]:vgpr(s96) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX10: G_STORE [[COPY1]](s96), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    %0:vgpr(p1) = COPY $vgpr0_vgpr1
    %1:vgpr(s96) = COPY $vgpr2_vgpr3_vgpr4
    G_STORE %1, %0 :: (store 12, align 16, addrspace 1)

...

---

name: store_global_v6s16
legalized:       true
tracksRegLiveness: true
regBankSelected: true

body: |
  bb.0:
    liveins:  $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4

    ; GFX6-LABEL: name: store_global_v6s16
    ; GFX6: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX6: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX6: [[COPY1:%[0-9]+]]:vgpr(<6 x s16>) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX6: G_STORE [[COPY1]](<6 x s16>), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX7-LABEL: name: store_global_v6s16
    ; GFX7: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX7: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX7: [[COPY1:%[0-9]+]]:vgpr(<6 x s16>) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX7: G_STORE [[COPY1]](<6 x s16>), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX7-FLAT-LABEL: name: store_global_v6s16
    ; GFX7-FLAT: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX7-FLAT: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX7-FLAT: [[COPY1:%[0-9]+]]:vgpr(<6 x s16>) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX7-FLAT: G_STORE [[COPY1]](<6 x s16>), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX8-LABEL: name: store_global_v6s16
    ; GFX8: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX8: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX8: [[COPY1:%[0-9]+]]:vgpr(<6 x s16>) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX8: G_STORE [[COPY1]](<6 x s16>), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX9-LABEL: name: store_global_v6s16
    ; GFX9: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX9: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX9: [[COPY1:%[0-9]+]]:vgpr(<6 x s16>) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX9: G_STORE [[COPY1]](<6 x s16>), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    ; GFX10-LABEL: name: store_global_v6s16
    ; GFX10: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3_vgpr4
    ; GFX10: [[COPY:%[0-9]+]]:vgpr(p1) = COPY $vgpr0_vgpr1
    ; GFX10: [[COPY1:%[0-9]+]]:vgpr(<6 x s16>) = COPY $vgpr2_vgpr3_vgpr4
    ; GFX10: G_STORE [[COPY1]](<6 x s16>), [[COPY]](p1) :: (store 12, align 16, addrspace 1)
    %0:vgpr(p1) = COPY $vgpr0_vgpr1
    %1:vgpr(<6 x s16>) = COPY $vgpr2_vgpr3_vgpr4
    G_STORE %1, %0 :: (store 12, align 16, addrspace 1)

...