forked from OSchip/llvm-project
493 lines
27 KiB
YAML
493 lines
27 KiB
YAML
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
|
|
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=hawaii -O0 -run-pass=legalizer %s -o - | FileCheck -check-prefix=GFX78 %s
|
|
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=fiji -O0 -run-pass=legalizer %s -o - | FileCheck -check-prefix=GFX78 %s
|
|
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -O0 -run-pass=legalizer %s -o - | FileCheck -check-prefix=GFX9 %s
|
|
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1010 -O0 -run-pass=legalizer %s -o - | FileCheck -check-prefix=GFX9 %s
|
|
|
|
---
|
|
name: build_vector_v2s16
|
|
body: |
|
|
bb.0:
|
|
liveins: $vgpr0, $vgpr1
|
|
|
|
; GFX78-LABEL: name: build_vector_v2s16
|
|
; GFX78: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX78: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX78: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
|
|
; GFX78: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
|
|
; GFX78: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
|
|
; GFX78: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
|
|
; GFX78: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
|
|
; GFX78: S_NOP 0, implicit [[BITCAST]](<2 x s16>)
|
|
; GFX9-LABEL: name: build_vector_v2s16
|
|
; GFX9: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX9: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX9: [[BUILD_VECTOR_TRUNC:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY]](s32), [[COPY1]](s32)
|
|
; GFX9: S_NOP 0, implicit [[BUILD_VECTOR_TRUNC]](<2 x s16>)
|
|
%0:_(s32) = COPY $vgpr0
|
|
%1:_(s32) = COPY $vgpr1
|
|
%2:_(s16) = G_TRUNC %0
|
|
%3:_(s16) = G_TRUNC %1
|
|
%4:_(<2 x s16>) = G_BUILD_VECTOR %2, %3
|
|
S_NOP 0, implicit %4
|
|
...
|
|
|
|
---
|
|
name: build_vector_v3s16
|
|
body: |
|
|
bb.0:
|
|
liveins: $vgpr0, $vgpr1, $vgpr2
|
|
|
|
; GFX78-LABEL: name: build_vector_v3s16
|
|
; GFX78: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX78: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX78: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX78: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
|
|
; GFX78: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
|
|
; GFX78: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
|
|
; GFX78: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
|
|
; GFX78: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
|
|
; GFX78: [[AND2:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
|
|
; GFX78: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[AND3]], [[C1]](s32)
|
|
; GFX78: [[OR1:%[0-9]+]]:_(s32) = G_OR [[AND2]], [[SHL1]]
|
|
; GFX78: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
|
|
; GFX78: [[AND4:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[AND5:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
|
|
; GFX78: [[SHL2:%[0-9]+]]:_(s32) = G_SHL [[AND5]], [[C1]](s32)
|
|
; GFX78: [[OR2:%[0-9]+]]:_(s32) = G_OR [[AND4]], [[SHL2]]
|
|
; GFX78: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR2]](s32)
|
|
; GFX78: [[CONCAT_VECTORS:%[0-9]+]]:_(<6 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>)
|
|
; GFX78: S_NOP 0, implicit [[CONCAT_VECTORS]](<6 x s16>)
|
|
; GFX9-LABEL: name: build_vector_v3s16
|
|
; GFX9: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX9: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX9: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX9: [[BUILD_VECTOR_TRUNC:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY]](s32), [[COPY1]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY2]](s32), [[COPY]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY1]](s32), [[COPY2]](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: S_NOP 0, implicit [[CONCAT_VECTORS]](<6 x s16>)
|
|
%0:_(s32) = COPY $vgpr0
|
|
%1:_(s32) = COPY $vgpr1
|
|
%2:_(s32) = COPY $vgpr2
|
|
%3:_(s16) = G_TRUNC %0
|
|
%4:_(s16) = G_TRUNC %1
|
|
%5:_(s16) = G_TRUNC %2
|
|
%6:_(<3 x s16>) = G_BUILD_VECTOR %3, %4, %5
|
|
%7:_(<6 x s16>) = G_CONCAT_VECTORS %6, %6
|
|
S_NOP 0, implicit %7
|
|
...
|
|
|
|
---
|
|
name: build_vector_v4s16
|
|
body: |
|
|
bb.0:
|
|
liveins: $vgpr0, $vgpr1, $vgpr2, $vgpr3
|
|
|
|
; GFX78-LABEL: name: build_vector_v4s16
|
|
; GFX78: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX78: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX78: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX78: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX78: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
|
|
; GFX78: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
|
|
; GFX78: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
|
|
; GFX78: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
|
|
; GFX78: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
|
|
; GFX78: [[AND2:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
|
|
; GFX78: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C]]
|
|
; GFX78: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[AND3]], [[C1]](s32)
|
|
; GFX78: [[OR1:%[0-9]+]]:_(s32) = G_OR [[AND2]], [[SHL1]]
|
|
; GFX78: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
|
|
; GFX78: [[CONCAT_VECTORS:%[0-9]+]]:_(<4 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>)
|
|
; GFX78: S_NOP 0, implicit [[CONCAT_VECTORS]](<4 x s16>)
|
|
; GFX9-LABEL: name: build_vector_v4s16
|
|
; GFX9: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX9: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX9: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX9: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX9: [[BUILD_VECTOR_TRUNC:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY]](s32), [[COPY1]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY2]](s32), [[COPY3]](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: S_NOP 0, implicit [[CONCAT_VECTORS]](<4 x s16>)
|
|
%0:_(s32) = COPY $vgpr0
|
|
%1:_(s32) = COPY $vgpr1
|
|
%2:_(s32) = COPY $vgpr2
|
|
%3:_(s32) = COPY $vgpr3
|
|
%4:_(s16) = G_TRUNC %0
|
|
%5:_(s16) = G_TRUNC %1
|
|
%6:_(s16) = G_TRUNC %2
|
|
%7:_(s16) = G_TRUNC %3
|
|
%8:_(<4 x s16>) = G_BUILD_VECTOR %4, %5, %6, %7
|
|
S_NOP 0, implicit %8
|
|
...
|
|
|
|
---
|
|
name: build_vector_v5s16
|
|
body: |
|
|
bb.0:
|
|
liveins: $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4
|
|
|
|
; GFX78-LABEL: name: build_vector_v5s16
|
|
; GFX78: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX78: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX78: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX78: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX78: [[COPY4:%[0-9]+]]:_(s32) = COPY $vgpr4
|
|
; GFX78: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
|
|
; GFX78: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
|
|
; GFX78: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
|
|
; GFX78: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
|
|
; GFX78: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
|
|
; GFX78: [[AND2:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
|
|
; GFX78: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C]]
|
|
; GFX78: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[AND3]], [[C1]](s32)
|
|
; GFX78: [[OR1:%[0-9]+]]:_(s32) = G_OR [[AND2]], [[SHL1]]
|
|
; GFX78: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
|
|
; GFX78: [[AND4:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C]]
|
|
; GFX78: [[AND5:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[SHL2:%[0-9]+]]:_(s32) = G_SHL [[AND5]], [[C1]](s32)
|
|
; GFX78: [[OR2:%[0-9]+]]:_(s32) = G_OR [[AND4]], [[SHL2]]
|
|
; GFX78: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR2]](s32)
|
|
; GFX78: [[AND6:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[AND7:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
|
|
; GFX78: [[SHL3:%[0-9]+]]:_(s32) = G_SHL [[AND7]], [[C1]](s32)
|
|
; GFX78: [[OR3:%[0-9]+]]:_(s32) = G_OR [[AND6]], [[SHL3]]
|
|
; GFX78: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR3]](s32)
|
|
; GFX78: [[AND8:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C]]
|
|
; GFX78: [[AND9:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C]]
|
|
; GFX78: [[SHL4:%[0-9]+]]:_(s32) = G_SHL [[AND9]], [[C1]](s32)
|
|
; GFX78: [[OR4:%[0-9]+]]:_(s32) = G_OR [[AND8]], [[SHL4]]
|
|
; GFX78: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR4]](s32)
|
|
; GFX78: [[CONCAT_VECTORS:%[0-9]+]]:_(<10 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>)
|
|
; GFX78: S_NOP 0, implicit [[CONCAT_VECTORS]](<10 x s16>)
|
|
; GFX9-LABEL: name: build_vector_v5s16
|
|
; GFX9: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX9: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX9: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX9: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX9: [[COPY4:%[0-9]+]]:_(s32) = COPY $vgpr4
|
|
; GFX9: [[BUILD_VECTOR_TRUNC:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY]](s32), [[COPY1]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY2]](s32), [[COPY3]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY4]](s32), [[COPY]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY1]](s32), [[COPY2]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY3]](s32), [[COPY4]](s32)
|
|
; GFX9: [[CONCAT_VECTORS:%[0-9]+]]:_(<10 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR_TRUNC]](<2 x s16>), [[BUILD_VECTOR_TRUNC1]](<2 x s16>), [[BUILD_VECTOR_TRUNC2]](<2 x s16>), [[BUILD_VECTOR_TRUNC3]](<2 x s16>), [[BUILD_VECTOR_TRUNC4]](<2 x s16>)
|
|
; GFX9: S_NOP 0, implicit [[CONCAT_VECTORS]](<10 x s16>)
|
|
%0:_(s32) = COPY $vgpr0
|
|
%1:_(s32) = COPY $vgpr1
|
|
%2:_(s32) = COPY $vgpr2
|
|
%3:_(s32) = COPY $vgpr3
|
|
%4:_(s32) = COPY $vgpr4
|
|
%5:_(s16) = G_TRUNC %0
|
|
%6:_(s16) = G_TRUNC %1
|
|
%7:_(s16) = G_TRUNC %2
|
|
%8:_(s16) = G_TRUNC %3
|
|
%9:_(s16) = G_TRUNC %4
|
|
%10:_(<5 x s16>) = G_BUILD_VECTOR %5, %6, %7, %8, %9
|
|
%11:_(<10 x s16>) = G_CONCAT_VECTORS %10, %10
|
|
S_NOP 0, implicit %11
|
|
...
|
|
|
|
---
|
|
name: build_vector_v7s16
|
|
body: |
|
|
bb.0:
|
|
liveins: $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6
|
|
|
|
; GFX78-LABEL: name: build_vector_v7s16
|
|
; GFX78: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX78: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX78: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX78: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX78: [[COPY4:%[0-9]+]]:_(s32) = COPY $vgpr4
|
|
; GFX78: [[COPY5:%[0-9]+]]:_(s32) = COPY $vgpr5
|
|
; GFX78: [[COPY6:%[0-9]+]]:_(s32) = COPY $vgpr6
|
|
; GFX78: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
|
|
; GFX78: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
|
|
; GFX78: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
|
|
; GFX78: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
|
|
; GFX78: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
|
|
; GFX78: [[AND2:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
|
|
; GFX78: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C]]
|
|
; GFX78: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[AND3]], [[C1]](s32)
|
|
; GFX78: [[OR1:%[0-9]+]]:_(s32) = G_OR [[AND2]], [[SHL1]]
|
|
; GFX78: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
|
|
; GFX78: [[AND4:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C]]
|
|
; GFX78: [[AND5:%[0-9]+]]:_(s32) = G_AND [[COPY5]], [[C]]
|
|
; GFX78: [[SHL2:%[0-9]+]]:_(s32) = G_SHL [[AND5]], [[C1]](s32)
|
|
; GFX78: [[OR2:%[0-9]+]]:_(s32) = G_OR [[AND4]], [[SHL2]]
|
|
; GFX78: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR2]](s32)
|
|
; GFX78: [[AND6:%[0-9]+]]:_(s32) = G_AND [[COPY6]], [[C]]
|
|
; GFX78: [[AND7:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[SHL3:%[0-9]+]]:_(s32) = G_SHL [[AND7]], [[C1]](s32)
|
|
; GFX78: [[OR3:%[0-9]+]]:_(s32) = G_OR [[AND6]], [[SHL3]]
|
|
; GFX78: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR3]](s32)
|
|
; GFX78: [[AND8:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[AND9:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
|
|
; GFX78: [[SHL4:%[0-9]+]]:_(s32) = G_SHL [[AND9]], [[C1]](s32)
|
|
; GFX78: [[OR4:%[0-9]+]]:_(s32) = G_OR [[AND8]], [[SHL4]]
|
|
; GFX78: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR4]](s32)
|
|
; GFX78: [[AND10:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C]]
|
|
; GFX78: [[AND11:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C]]
|
|
; GFX78: [[SHL5:%[0-9]+]]:_(s32) = G_SHL [[AND11]], [[C1]](s32)
|
|
; GFX78: [[OR5:%[0-9]+]]:_(s32) = G_OR [[AND10]], [[SHL5]]
|
|
; GFX78: [[BITCAST5:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR5]](s32)
|
|
; GFX78: [[AND12:%[0-9]+]]:_(s32) = G_AND [[COPY5]], [[C]]
|
|
; GFX78: [[AND13:%[0-9]+]]:_(s32) = G_AND [[COPY6]], [[C]]
|
|
; GFX78: [[SHL6:%[0-9]+]]:_(s32) = G_SHL [[AND13]], [[C1]](s32)
|
|
; GFX78: [[OR6:%[0-9]+]]:_(s32) = G_OR [[AND12]], [[SHL6]]
|
|
; GFX78: [[BITCAST6:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR6]](s32)
|
|
; GFX78: [[CONCAT_VECTORS:%[0-9]+]]:_(<14 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>), [[BITCAST5]](<2 x s16>), [[BITCAST6]](<2 x s16>)
|
|
; GFX78: S_NOP 0, implicit [[CONCAT_VECTORS]](<14 x s16>)
|
|
; GFX9-LABEL: name: build_vector_v7s16
|
|
; GFX9: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX9: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX9: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX9: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX9: [[COPY4:%[0-9]+]]:_(s32) = COPY $vgpr4
|
|
; GFX9: [[COPY5:%[0-9]+]]:_(s32) = COPY $vgpr5
|
|
; GFX9: [[COPY6:%[0-9]+]]:_(s32) = COPY $vgpr6
|
|
; GFX9: [[BUILD_VECTOR_TRUNC:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY]](s32), [[COPY1]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY2]](s32), [[COPY3]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY4]](s32), [[COPY5]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY6]](s32), [[COPY]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY1]](s32), [[COPY2]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY3]](s32), [[COPY4]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY5]](s32), [[COPY6]](s32)
|
|
; GFX9: [[CONCAT_VECTORS:%[0-9]+]]:_(<14 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR_TRUNC]](<2 x s16>), [[BUILD_VECTOR_TRUNC1]](<2 x s16>), [[BUILD_VECTOR_TRUNC2]](<2 x s16>), [[BUILD_VECTOR_TRUNC3]](<2 x s16>), [[BUILD_VECTOR_TRUNC4]](<2 x s16>), [[BUILD_VECTOR_TRUNC5]](<2 x s16>), [[BUILD_VECTOR_TRUNC6]](<2 x s16>)
|
|
; GFX9: S_NOP 0, implicit [[CONCAT_VECTORS]](<14 x s16>)
|
|
%0:_(s32) = COPY $vgpr0
|
|
%1:_(s32) = COPY $vgpr1
|
|
%2:_(s32) = COPY $vgpr2
|
|
%3:_(s32) = COPY $vgpr3
|
|
%4:_(s32) = COPY $vgpr4
|
|
%5:_(s32) = COPY $vgpr5
|
|
%6:_(s32) = COPY $vgpr6
|
|
%7:_(s16) = G_TRUNC %0
|
|
%8:_(s16) = G_TRUNC %1
|
|
%9:_(s16) = G_TRUNC %2
|
|
%10:_(s16) = G_TRUNC %3
|
|
%11:_(s16) = G_TRUNC %4
|
|
%12:_(s16) = G_TRUNC %5
|
|
%13:_(s16) = G_TRUNC %6
|
|
%14:_(<7 x s16>) = G_BUILD_VECTOR %7, %8, %9, %10, %11, %12, %13
|
|
%15:_(<14 x s16>) = G_CONCAT_VECTORS %14, %14
|
|
S_NOP 0, implicit %15
|
|
...
|
|
|
|
---
|
|
name: build_vector_v8s16
|
|
body: |
|
|
bb.0:
|
|
liveins: $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7
|
|
|
|
; GFX78-LABEL: name: build_vector_v8s16
|
|
; GFX78: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX78: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX78: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX78: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX78: [[COPY4:%[0-9]+]]:_(s32) = COPY $vgpr4
|
|
; GFX78: [[COPY5:%[0-9]+]]:_(s32) = COPY $vgpr5
|
|
; GFX78: [[COPY6:%[0-9]+]]:_(s32) = COPY $vgpr6
|
|
; GFX78: [[COPY7:%[0-9]+]]:_(s32) = COPY $vgpr7
|
|
; GFX78: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
|
|
; GFX78: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
|
|
; GFX78: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
|
|
; GFX78: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
|
|
; GFX78: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
|
|
; GFX78: [[AND2:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
|
|
; GFX78: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C]]
|
|
; GFX78: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[AND3]], [[C1]](s32)
|
|
; GFX78: [[OR1:%[0-9]+]]:_(s32) = G_OR [[AND2]], [[SHL1]]
|
|
; GFX78: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
|
|
; GFX78: [[AND4:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C]]
|
|
; GFX78: [[AND5:%[0-9]+]]:_(s32) = G_AND [[COPY5]], [[C]]
|
|
; GFX78: [[SHL2:%[0-9]+]]:_(s32) = G_SHL [[AND5]], [[C1]](s32)
|
|
; GFX78: [[OR2:%[0-9]+]]:_(s32) = G_OR [[AND4]], [[SHL2]]
|
|
; GFX78: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR2]](s32)
|
|
; GFX78: [[AND6:%[0-9]+]]:_(s32) = G_AND [[COPY6]], [[C]]
|
|
; GFX78: [[AND7:%[0-9]+]]:_(s32) = G_AND [[COPY7]], [[C]]
|
|
; GFX78: [[SHL3:%[0-9]+]]:_(s32) = G_SHL [[AND7]], [[C1]](s32)
|
|
; GFX78: [[OR3:%[0-9]+]]:_(s32) = G_OR [[AND6]], [[SHL3]]
|
|
; GFX78: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR3]](s32)
|
|
; GFX78: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>)
|
|
; GFX78: S_NOP 0, implicit [[CONCAT_VECTORS]](<8 x s16>)
|
|
; GFX9-LABEL: name: build_vector_v8s16
|
|
; GFX9: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX9: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX9: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX9: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX9: [[COPY4:%[0-9]+]]:_(s32) = COPY $vgpr4
|
|
; GFX9: [[COPY5:%[0-9]+]]:_(s32) = COPY $vgpr5
|
|
; GFX9: [[COPY6:%[0-9]+]]:_(s32) = COPY $vgpr6
|
|
; GFX9: [[COPY7:%[0-9]+]]:_(s32) = COPY $vgpr7
|
|
; GFX9: [[BUILD_VECTOR_TRUNC:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY]](s32), [[COPY1]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY2]](s32), [[COPY3]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY4]](s32), [[COPY5]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY6]](s32), [[COPY7]](s32)
|
|
; GFX9: [[CONCAT_VECTORS:%[0-9]+]]:_(<8 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR_TRUNC]](<2 x s16>), [[BUILD_VECTOR_TRUNC1]](<2 x s16>), [[BUILD_VECTOR_TRUNC2]](<2 x s16>), [[BUILD_VECTOR_TRUNC3]](<2 x s16>)
|
|
; GFX9: S_NOP 0, implicit [[CONCAT_VECTORS]](<8 x s16>)
|
|
%0:_(s32) = COPY $vgpr0
|
|
%1:_(s32) = COPY $vgpr1
|
|
%2:_(s32) = COPY $vgpr2
|
|
%3:_(s32) = COPY $vgpr3
|
|
%4:_(s32) = COPY $vgpr4
|
|
%5:_(s32) = COPY $vgpr5
|
|
%6:_(s32) = COPY $vgpr6
|
|
%7:_(s32) = COPY $vgpr7
|
|
%8:_(s16) = G_TRUNC %0
|
|
%9:_(s16) = G_TRUNC %1
|
|
%10:_(s16) = G_TRUNC %2
|
|
%11:_(s16) = G_TRUNC %3
|
|
%12:_(s16) = G_TRUNC %4
|
|
%13:_(s16) = G_TRUNC %5
|
|
%14:_(s16) = G_TRUNC %6
|
|
%15:_(s16) = G_TRUNC %7
|
|
%16:_(<8 x s16>) = G_BUILD_VECTOR %8, %9, %10, %11, %12, %13, %14, %15
|
|
S_NOP 0, implicit %16
|
|
...
|
|
|
|
---
|
|
name: build_vector_v16s16
|
|
body: |
|
|
bb.0:
|
|
liveins: $vgpr0, $vgpr1, $vgpr2, $vgpr3, $vgpr4, $vgpr5, $vgpr6, $vgpr7, $vgpr8, $vgpr9, $vgpr10, $vgpr11, $vgpr12, $vgpr13, $vgpr14, $vgpr15
|
|
|
|
; GFX78-LABEL: name: build_vector_v16s16
|
|
; GFX78: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX78: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX78: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX78: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX78: [[COPY4:%[0-9]+]]:_(s32) = COPY $vgpr4
|
|
; GFX78: [[COPY5:%[0-9]+]]:_(s32) = COPY $vgpr5
|
|
; GFX78: [[COPY6:%[0-9]+]]:_(s32) = COPY $vgpr6
|
|
; GFX78: [[COPY7:%[0-9]+]]:_(s32) = COPY $vgpr7
|
|
; GFX78: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr8
|
|
; GFX78: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr9
|
|
; GFX78: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr10
|
|
; GFX78: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr11
|
|
; GFX78: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr12
|
|
; GFX78: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr13
|
|
; GFX78: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr14
|
|
; GFX78: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr15
|
|
; GFX78: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 65535
|
|
; GFX78: [[AND:%[0-9]+]]:_(s32) = G_AND [[COPY]], [[C]]
|
|
; GFX78: [[AND1:%[0-9]+]]:_(s32) = G_AND [[COPY1]], [[C]]
|
|
; GFX78: [[C1:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
|
|
; GFX78: [[SHL:%[0-9]+]]:_(s32) = G_SHL [[AND1]], [[C1]](s32)
|
|
; GFX78: [[OR:%[0-9]+]]:_(s32) = G_OR [[AND]], [[SHL]]
|
|
; GFX78: [[BITCAST:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR]](s32)
|
|
; GFX78: [[AND2:%[0-9]+]]:_(s32) = G_AND [[COPY2]], [[C]]
|
|
; GFX78: [[AND3:%[0-9]+]]:_(s32) = G_AND [[COPY3]], [[C]]
|
|
; GFX78: [[SHL1:%[0-9]+]]:_(s32) = G_SHL [[AND3]], [[C1]](s32)
|
|
; GFX78: [[OR1:%[0-9]+]]:_(s32) = G_OR [[AND2]], [[SHL1]]
|
|
; GFX78: [[BITCAST1:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR1]](s32)
|
|
; GFX78: [[AND4:%[0-9]+]]:_(s32) = G_AND [[COPY4]], [[C]]
|
|
; GFX78: [[AND5:%[0-9]+]]:_(s32) = G_AND [[COPY5]], [[C]]
|
|
; GFX78: [[SHL2:%[0-9]+]]:_(s32) = G_SHL [[AND5]], [[C1]](s32)
|
|
; GFX78: [[OR2:%[0-9]+]]:_(s32) = G_OR [[AND4]], [[SHL2]]
|
|
; GFX78: [[BITCAST2:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR2]](s32)
|
|
; GFX78: [[AND6:%[0-9]+]]:_(s32) = G_AND [[COPY6]], [[C]]
|
|
; GFX78: [[AND7:%[0-9]+]]:_(s32) = G_AND [[COPY7]], [[C]]
|
|
; GFX78: [[SHL3:%[0-9]+]]:_(s32) = G_SHL [[AND7]], [[C1]](s32)
|
|
; GFX78: [[OR3:%[0-9]+]]:_(s32) = G_OR [[AND6]], [[SHL3]]
|
|
; GFX78: [[BITCAST3:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR3]](s32)
|
|
; GFX78: [[AND8:%[0-9]+]]:_(s32) = G_AND [[COPY8]], [[C]]
|
|
; GFX78: [[AND9:%[0-9]+]]:_(s32) = G_AND [[COPY9]], [[C]]
|
|
; GFX78: [[SHL4:%[0-9]+]]:_(s32) = G_SHL [[AND9]], [[C1]](s32)
|
|
; GFX78: [[OR4:%[0-9]+]]:_(s32) = G_OR [[AND8]], [[SHL4]]
|
|
; GFX78: [[BITCAST4:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR4]](s32)
|
|
; GFX78: [[AND10:%[0-9]+]]:_(s32) = G_AND [[COPY10]], [[C]]
|
|
; GFX78: [[AND11:%[0-9]+]]:_(s32) = G_AND [[COPY11]], [[C]]
|
|
; GFX78: [[SHL5:%[0-9]+]]:_(s32) = G_SHL [[AND11]], [[C1]](s32)
|
|
; GFX78: [[OR5:%[0-9]+]]:_(s32) = G_OR [[AND10]], [[SHL5]]
|
|
; GFX78: [[BITCAST5:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR5]](s32)
|
|
; GFX78: [[AND12:%[0-9]+]]:_(s32) = G_AND [[COPY12]], [[C]]
|
|
; GFX78: [[AND13:%[0-9]+]]:_(s32) = G_AND [[COPY13]], [[C]]
|
|
; GFX78: [[SHL6:%[0-9]+]]:_(s32) = G_SHL [[AND13]], [[C1]](s32)
|
|
; GFX78: [[OR6:%[0-9]+]]:_(s32) = G_OR [[AND12]], [[SHL6]]
|
|
; GFX78: [[BITCAST6:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR6]](s32)
|
|
; GFX78: [[AND14:%[0-9]+]]:_(s32) = G_AND [[COPY14]], [[C]]
|
|
; GFX78: [[AND15:%[0-9]+]]:_(s32) = G_AND [[COPY15]], [[C]]
|
|
; GFX78: [[SHL7:%[0-9]+]]:_(s32) = G_SHL [[AND15]], [[C1]](s32)
|
|
; GFX78: [[OR7:%[0-9]+]]:_(s32) = G_OR [[AND14]], [[SHL7]]
|
|
; GFX78: [[BITCAST7:%[0-9]+]]:_(<2 x s16>) = G_BITCAST [[OR7]](s32)
|
|
; GFX78: [[CONCAT_VECTORS:%[0-9]+]]:_(<16 x s16>) = G_CONCAT_VECTORS [[BITCAST]](<2 x s16>), [[BITCAST1]](<2 x s16>), [[BITCAST2]](<2 x s16>), [[BITCAST3]](<2 x s16>), [[BITCAST4]](<2 x s16>), [[BITCAST5]](<2 x s16>), [[BITCAST6]](<2 x s16>), [[BITCAST7]](<2 x s16>)
|
|
; GFX78: S_NOP 0, implicit [[CONCAT_VECTORS]](<16 x s16>)
|
|
; GFX9-LABEL: name: build_vector_v16s16
|
|
; GFX9: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
|
|
; GFX9: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
|
|
; GFX9: [[COPY2:%[0-9]+]]:_(s32) = COPY $vgpr2
|
|
; GFX9: [[COPY3:%[0-9]+]]:_(s32) = COPY $vgpr3
|
|
; GFX9: [[COPY4:%[0-9]+]]:_(s32) = COPY $vgpr4
|
|
; GFX9: [[COPY5:%[0-9]+]]:_(s32) = COPY $vgpr5
|
|
; GFX9: [[COPY6:%[0-9]+]]:_(s32) = COPY $vgpr6
|
|
; GFX9: [[COPY7:%[0-9]+]]:_(s32) = COPY $vgpr7
|
|
; GFX9: [[COPY8:%[0-9]+]]:_(s32) = COPY $vgpr8
|
|
; GFX9: [[COPY9:%[0-9]+]]:_(s32) = COPY $vgpr9
|
|
; GFX9: [[COPY10:%[0-9]+]]:_(s32) = COPY $vgpr10
|
|
; GFX9: [[COPY11:%[0-9]+]]:_(s32) = COPY $vgpr11
|
|
; GFX9: [[COPY12:%[0-9]+]]:_(s32) = COPY $vgpr12
|
|
; GFX9: [[COPY13:%[0-9]+]]:_(s32) = COPY $vgpr13
|
|
; GFX9: [[COPY14:%[0-9]+]]:_(s32) = COPY $vgpr14
|
|
; GFX9: [[COPY15:%[0-9]+]]:_(s32) = COPY $vgpr15
|
|
; GFX9: [[BUILD_VECTOR_TRUNC:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY]](s32), [[COPY1]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC1:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY2]](s32), [[COPY3]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC2:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY4]](s32), [[COPY5]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC3:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY6]](s32), [[COPY7]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC4:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY8]](s32), [[COPY9]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC5:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY10]](s32), [[COPY11]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC6:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY12]](s32), [[COPY13]](s32)
|
|
; GFX9: [[BUILD_VECTOR_TRUNC7:%[0-9]+]]:_(<2 x s16>) = G_BUILD_VECTOR_TRUNC [[COPY14]](s32), [[COPY15]](s32)
|
|
; GFX9: [[CONCAT_VECTORS:%[0-9]+]]:_(<16 x s16>) = G_CONCAT_VECTORS [[BUILD_VECTOR_TRUNC]](<2 x s16>), [[BUILD_VECTOR_TRUNC1]](<2 x s16>), [[BUILD_VECTOR_TRUNC2]](<2 x s16>), [[BUILD_VECTOR_TRUNC3]](<2 x s16>), [[BUILD_VECTOR_TRUNC4]](<2 x s16>), [[BUILD_VECTOR_TRUNC5]](<2 x s16>), [[BUILD_VECTOR_TRUNC6]](<2 x s16>), [[BUILD_VECTOR_TRUNC7]](<2 x s16>)
|
|
; GFX9: S_NOP 0, implicit [[CONCAT_VECTORS]](<16 x s16>)
|
|
%0:_(s32) = COPY $vgpr0
|
|
%1:_(s32) = COPY $vgpr1
|
|
%2:_(s32) = COPY $vgpr2
|
|
%3:_(s32) = COPY $vgpr3
|
|
%4:_(s32) = COPY $vgpr4
|
|
%5:_(s32) = COPY $vgpr5
|
|
%6:_(s32) = COPY $vgpr6
|
|
%7:_(s32) = COPY $vgpr7
|
|
%8:_(s32) = COPY $vgpr8
|
|
%9:_(s32) = COPY $vgpr9
|
|
%10:_(s32) = COPY $vgpr10
|
|
%11:_(s32) = COPY $vgpr11
|
|
%12:_(s32) = COPY $vgpr12
|
|
%13:_(s32) = COPY $vgpr13
|
|
%14:_(s32) = COPY $vgpr14
|
|
%15:_(s32) = COPY $vgpr15
|
|
%16:_(s16) = G_TRUNC %0
|
|
%17:_(s16) = G_TRUNC %1
|
|
%18:_(s16) = G_TRUNC %2
|
|
%19:_(s16) = G_TRUNC %3
|
|
%20:_(s16) = G_TRUNC %4
|
|
%21:_(s16) = G_TRUNC %5
|
|
%22:_(s16) = G_TRUNC %6
|
|
%23:_(s16) = G_TRUNC %7
|
|
%24:_(s16) = G_TRUNC %8
|
|
%25:_(s16) = G_TRUNC %9
|
|
%26:_(s16) = G_TRUNC %10
|
|
%27:_(s16) = G_TRUNC %11
|
|
%28:_(s16) = G_TRUNC %12
|
|
%29:_(s16) = G_TRUNC %13
|
|
%30:_(s16) = G_TRUNC %14
|
|
%31:_(s16) = G_TRUNC %15
|
|
%32:_(<16 x s16>) = G_BUILD_VECTOR %16, %17, %18, %19, %20, %21, %22, %23, %24, %25, %26, %27, %28, %29, %30, %31
|
|
S_NOP 0, implicit %32
|
|
...
|