-
Notifications
You must be signed in to change notification settings - Fork 14.9k
[AMDGPU] gfx1250 min/max codegen tests. NFC. #155292
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[AMDGPU] gfx1250 min/max codegen tests. NFC. #155292
Conversation
@llvm/pr-subscribers-llvm-globalisel Author: Stanislav Mekhanoshin (rampitec) ChangesPatch is 80.26 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/155292.diff 10 Files Affected:
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smax-64.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smax-64.mir
new file mode 100644
index 0000000000000..ace459979833f
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smax-64.mir
@@ -0,0 +1,65 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
+
+---
+name: smax_s64_sv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: smax_s64_sv
+ ; GCN: liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MAX_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_I64_e64_]]
+ %0:sgpr(s64) = COPY $sgpr0_sgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_SMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: smax_s64_vs
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-LABEL: name: smax_s64_vs
+ ; GCN: liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $sgpr2_sgpr3
+ ; GCN-NEXT: [[V_MAX_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_I64_e64_]]
+ %0:sgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $sgpr2_sgpr3
+ %2:vgpr(s64) = G_SMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: smax_s64_vv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: smax_s64_vv
+ ; GCN: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MAX_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_I64_e64_]]
+ %0:vgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_SMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin-64.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin-64.mir
new file mode 100644
index 0000000000000..f341bdfb22ab7
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin-64.mir
@@ -0,0 +1,65 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
+
+---
+name: smin_s64_sv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: smin_s64_sv
+ ; GCN: liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MIN_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_I64_e64_]]
+ %0:sgpr(s64) = COPY $sgpr0_sgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_SMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: smin_s64_vs
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-LABEL: name: smin_s64_vs
+ ; GCN: liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $sgpr2_sgpr3
+ ; GCN-NEXT: [[V_MIN_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_I64_e64_]]
+ %0:sgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $sgpr2_sgpr3
+ %2:vgpr(s64) = G_SMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: smin_s64_vv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: smin_s64_vv
+ ; GCN: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MIN_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_I64_e64_]]
+ %0:vgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_SMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin.mir
index 33f14c179f2a9..2df27bdd459d5 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin.mir
@@ -2,6 +2,7 @@
# RUN: llc -mtriple=amdgcn -mcpu=gfx900 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefix=GCN %s
# RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefix=GCN %s
# RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefix=GCN %s
+# RUN: llc -mtriple=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
---
name: smin_s32_ss
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umax-64.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umax-64.mir
new file mode 100644
index 0000000000000..9edcf573c8332
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umax-64.mir
@@ -0,0 +1,65 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
+
+---
+name: umax_s64_sv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: umax_s64_sv
+ ; GCN: liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MAX_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_U64_e64_]]
+ %0:sgpr(s64) = COPY $sgpr0_sgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_UMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: umax_s64_vs
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-LABEL: name: umax_s64_vs
+ ; GCN: liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $sgpr2_sgpr3
+ ; GCN-NEXT: [[V_MAX_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_U64_e64_]]
+ %0:sgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $sgpr2_sgpr3
+ %2:vgpr(s64) = G_UMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: umax_s64_vv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: umax_s64_vv
+ ; GCN: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MAX_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_U64_e64_]]
+ %0:vgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_UMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umin-64.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umin-64.mir
new file mode 100644
index 0000000000000..e6c68112d067f
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umin-64.mir
@@ -0,0 +1,65 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
+
+---
+name: umin_s64_sv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: umin_s64_sv
+ ; GCN: liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MIN_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_U64_e64_]]
+ %0:sgpr(s64) = COPY $sgpr0_sgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_UMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: umin_s64_vs
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-LABEL: name: umin_s64_vs
+ ; GCN: liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $sgpr2_sgpr3
+ ; GCN-NEXT: [[V_MIN_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_U64_e64_]]
+ %0:sgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $sgpr2_sgpr3
+ %2:vgpr(s64) = G_UMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: umin_s64_vv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: umin_s64_vv
+ ; GCN: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MIN_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_U64_e64_]]
+ %0:vgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_UMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-smax.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-smax.mir
index db11855d2967b..45714fd99d7bd 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-smax.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-smax.mir
@@ -4,6 +4,7 @@
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -run-pass=legalizer %s -o - | FileCheck -check-prefix=GFX9 %s
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -run-pass=legalizer %s -o - | FileCheck -check-prefix=GFX9 %s
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -mattr=-real-true16 -run-pass=legalizer %s -o - | FileCheck -check-prefixes=GFX9 %s
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1250 -mattr=-real-true16 -run-pass=legalizer %s -o - | FileCheck -check-prefixes=GFX1250 %s
---
name: test_smax_s32
@@ -34,6 +35,14 @@ body: |
; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[COPY]], [[COPY1]]
; GFX9-NEXT: $vgpr0 = COPY [[SMAX]](s32)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s32
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[COPY]], [[COPY1]]
+ ; GFX1250-NEXT: $vgpr0 = COPY [[SMAX]](s32)
%0:_(s32) = COPY $vgpr0
%1:_(s32) = COPY $vgpr1
%2:_(s32) = G_SMAX %0, %1
@@ -72,6 +81,14 @@ body: |
; GFX9-NEXT: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(sgt), [[COPY]](s64), [[COPY1]]
; GFX9-NEXT: [[SELECT:%[0-9]+]]:_(s64) = G_SELECT [[ICMP]](s1), [[COPY]], [[COPY1]]
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[SELECT]](s64)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s64
+ ; GFX1250: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s64) = COPY $vgpr2_vgpr3
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s64) = G_SMAX [[COPY]], [[COPY1]]
+ ; GFX1250-NEXT: $vgpr0_vgpr1 = COPY [[SMAX]](s64)
%0:_(s64) = COPY $vgpr0_vgpr1
%1:_(s64) = COPY $vgpr2_vgpr3
%2:_(s64) = G_SMAX %0, %1
@@ -115,6 +132,17 @@ body: |
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(s16) = G_SMAX [[TRUNC]], [[TRUNC1]]
; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX]](s16)
; GFX9-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s16
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+ ; GFX1250-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s32)
+ ; GFX1250-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY1]](s32)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s16) = G_SMAX [[TRUNC]], [[TRUNC1]]
+ ; GFX1250-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX]](s16)
+ ; GFX1250-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
%0:_(s32) = COPY $vgpr0
%1:_(s32) = COPY $vgpr1
%2:_(s16) = G_TRUNC %0
@@ -165,6 +193,19 @@ body: |
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(s16) = G_SMAX [[TRUNC]], [[TRUNC1]]
; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX]](s16)
; GFX9-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s8
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+ ; GFX1250-NEXT: [[SEXT_INREG:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY]], 8
+ ; GFX1250-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[SEXT_INREG]](s32)
+ ; GFX1250-NEXT: [[SEXT_INREG1:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY1]], 8
+ ; GFX1250-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[SEXT_INREG1]](s32)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s16) = G_SMAX [[TRUNC]], [[TRUNC1]]
+ ; GFX1250-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX]](s16)
+ ; GFX1250-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
%0:_(s32) = COPY $vgpr0
%1:_(s32) = COPY $vgpr1
%2:_(s8) = G_TRUNC %0
@@ -209,6 +250,16 @@ body: |
; GFX9-NEXT: [[SEXT_INREG1:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY1]], 17
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[SEXT_INREG]], [[SEXT_INREG1]]
; GFX9-NEXT: $vgpr0 = COPY [[SMAX]](s32)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s17
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+ ; GFX1250-NEXT: [[SEXT_INREG:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY]], 17
+ ; GFX1250-NEXT: [[SEXT_INREG1:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY1]], 17
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[SEXT_INREG]], [[SEXT_INREG1]]
+ ; GFX1250-NEXT: $vgpr0 = COPY [[SMAX]](s32)
%0:_(s32) = COPY $vgpr0
%1:_(s32) = COPY $vgpr1
%2:_(s17) = G_TRUNC %0
@@ -259,6 +310,18 @@ body: |
; GFX9-NEXT: [[SMAX1:%[0-9]+]]:_(s32) = G_SMAX [[UV1]], [[UV3]]
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[SMAX]](s32), [[SMAX1]](s32)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
+ ;
+ ; GFX1250-LABEL: name: test_smax_v2s32
+ ; GFX1250: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
+ ; GFX1250-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
+ ; GFX1250-NEXT: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<2 x s32>)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[UV]], [[UV2]]
+ ; GFX1250-NEXT: [[SMAX1:%[0-9]+]]:_(s32) = G_SMAX [[UV1]], [[UV3]]
+ ; GFX1250-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[SMAX]](s32), [[SMAX1]](s32)
+ ; GFX1250-NEXT: $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_SMAX %0, %1
@@ -309,6 +372,19 @@ body: |
; GFX9-NEXT: [[SMAX2:%[0-9]+]]:_(s32) = G_SMAX [[UV2]], [[UV5]]
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[SMAX]](s32), [[SMAX1]](s32), [[SMAX2]](s32)
; GFX9-NEXT: $vgpr0_vgpr1_vgpr2 = COPY [[BUILD_VECTOR]](<3 x s32>)
+ ;
+ ; GFX1250-LABEL: name: test_smax_v3s32
+ ; GFX1250: liveins: $vgpr0_vgpr1_vgpr2, $vgpr3_vgpr4_vgpr5
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr3_vgpr4_vgpr5
+ ; GFX1250-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<3 x s32>)
+ ; GFX1250-NEXT: [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<3 x s32>)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[UV]], [[UV3]]
+ ; GFX1250-NEXT: [[SMAX1:%[0-9]+]]:_(s32) = G_SMAX [[UV1]], [[UV4]]
+ ; GFX1250-NEXT: [[SMAX2:%[0-9]+]]:_(s32) = G_SMAX [[UV2]], [[UV5]]
+ ; GFX1250-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[SMAX]](s32), [[SMAX1]](s32), [[SMAX2]](s32)
+ ; GFX1250-NEXT: $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_SMAX %0, %1
@@ -375,6 +451,14 @@ body: |
; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(<2 x s16>) = G_SMAX [[COPY]], [[COPY1]]
; GFX9-NEXT: $vgpr0 = COPY [[SMAX]](<2 x s16>)
+ ;
+ ; GFX1250-LABEL: name: test_smax_v2s16
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(<2 x s16>) = G_SMAX [[COPY]], [[COPY1]]
+ ; GFX1250-NEXT: $vgpr0 = COPY [[SMAX]](<2 x s16>)
%0:_(<2 x s16>) = COPY $vgpr0
%1:_(<2 x s16>) = COPY $vgpr1
%2:_(<2 x s16>) = G_SMAX %0, %1
@@ -461,6 +545,26 @@ body: |
; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[BITCAST2]](s32), [[LSHR]](s32), [[ANYEXT]](s32)
; GFX9-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<3 x s32>)
+ ;
+ ; GFX1250-LABEL: name: test_smax_v3s16
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[DEF:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
+ ; GFX1250-NEXT: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[DEF]](<4 x s16>)
+ ; GFX1250-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>)
+ ; GFX1250-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
+ ; GFX1250-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
+ ; GFX1250-NEXT: [[DEF1:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
+ ; GFX1250-NEXT: [[UV2:%[0-9]+]]:_(<2 x s16>), [[UV3:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[DEF1]](<4 x s16>)
+ ; GFX1250-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[UV3]](<2 x s16>)
+ ; GFX1250-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(<2 x s16>...
[truncated]
|
@llvm/pr-subscribers-backend-amdgpu Author: Stanislav Mekhanoshin (rampitec) ChangesPatch is 80.26 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/155292.diff 10 Files Affected:
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smax-64.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smax-64.mir
new file mode 100644
index 0000000000000..ace459979833f
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smax-64.mir
@@ -0,0 +1,65 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
+
+---
+name: smax_s64_sv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: smax_s64_sv
+ ; GCN: liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MAX_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_I64_e64_]]
+ %0:sgpr(s64) = COPY $sgpr0_sgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_SMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: smax_s64_vs
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-LABEL: name: smax_s64_vs
+ ; GCN: liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $sgpr2_sgpr3
+ ; GCN-NEXT: [[V_MAX_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_I64_e64_]]
+ %0:sgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $sgpr2_sgpr3
+ %2:vgpr(s64) = G_SMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: smax_s64_vv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: smax_s64_vv
+ ; GCN: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MAX_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_I64_e64_]]
+ %0:vgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_SMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin-64.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin-64.mir
new file mode 100644
index 0000000000000..f341bdfb22ab7
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin-64.mir
@@ -0,0 +1,65 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
+
+---
+name: smin_s64_sv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: smin_s64_sv
+ ; GCN: liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MIN_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_I64_e64_]]
+ %0:sgpr(s64) = COPY $sgpr0_sgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_SMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: smin_s64_vs
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-LABEL: name: smin_s64_vs
+ ; GCN: liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $sgpr2_sgpr3
+ ; GCN-NEXT: [[V_MIN_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_I64_e64_]]
+ %0:sgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $sgpr2_sgpr3
+ %2:vgpr(s64) = G_SMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: smin_s64_vv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: smin_s64_vv
+ ; GCN: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MIN_I64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_I64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_I64_e64_]]
+ %0:vgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_SMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin.mir
index 33f14c179f2a9..2df27bdd459d5 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-smin.mir
@@ -2,6 +2,7 @@
# RUN: llc -mtriple=amdgcn -mcpu=gfx900 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefix=GCN %s
# RUN: llc -mtriple=amdgcn -mcpu=gfx1010 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefix=GCN %s
# RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefix=GCN %s
+# RUN: llc -mtriple=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
---
name: smin_s32_ss
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umax-64.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umax-64.mir
new file mode 100644
index 0000000000000..9edcf573c8332
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umax-64.mir
@@ -0,0 +1,65 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
+
+---
+name: umax_s64_sv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: umax_s64_sv
+ ; GCN: liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MAX_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_U64_e64_]]
+ %0:sgpr(s64) = COPY $sgpr0_sgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_UMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: umax_s64_vs
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-LABEL: name: umax_s64_vs
+ ; GCN: liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $sgpr2_sgpr3
+ ; GCN-NEXT: [[V_MAX_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_U64_e64_]]
+ %0:sgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $sgpr2_sgpr3
+ %2:vgpr(s64) = G_UMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: umax_s64_vv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: umax_s64_vv
+ ; GCN: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MAX_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MAX_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MAX_U64_e64_]]
+ %0:vgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_UMAX %0, %1
+ S_ENDPGM 0, implicit %2
+...
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umin-64.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umin-64.mir
new file mode 100644
index 0000000000000..e6c68112d067f
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-umin-64.mir
@@ -0,0 +1,65 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
+# RUN: llc -march=amdgcn -mcpu=gfx1250 -run-pass=instruction-select %s -o - | FileCheck -check-prefix=GCN %s
+
+---
+name: umin_s64_sv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: umin_s64_sv
+ ; GCN: liveins: $sgpr0_sgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $sgpr0_sgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MIN_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_U64_e64_]]
+ %0:sgpr(s64) = COPY $sgpr0_sgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_UMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: umin_s64_vs
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-LABEL: name: umin_s64_vs
+ ; GCN: liveins: $vgpr0_vgpr1, $sgpr2_sgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:sreg_64 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $sgpr2_sgpr3
+ ; GCN-NEXT: [[V_MIN_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_U64_e64_]]
+ %0:sgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $sgpr2_sgpr3
+ %2:vgpr(s64) = G_UMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
+
+---
+name: umin_s64_vv
+legalized: true
+regBankSelected: true
+
+body: |
+ bb.0:
+ liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-LABEL: name: umin_s64_vv
+ ; GCN: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GCN-NEXT: {{ $}}
+ ; GCN-NEXT: [[COPY:%[0-9]+]]:vreg_64_align2 = COPY $vgpr0_vgpr1
+ ; GCN-NEXT: [[COPY1:%[0-9]+]]:vreg_64_align2 = COPY $vgpr2_vgpr3
+ ; GCN-NEXT: [[V_MIN_U64_e64_:%[0-9]+]]:vreg_64_align2 = V_MIN_U64_e64 [[COPY]], [[COPY1]], implicit $exec
+ ; GCN-NEXT: S_ENDPGM 0, implicit [[V_MIN_U64_e64_]]
+ %0:vgpr(s64) = COPY $vgpr0_vgpr1
+ %1:vgpr(s64) = COPY $vgpr2_vgpr3
+ %2:vgpr(s64) = G_UMIN %0, %1
+ S_ENDPGM 0, implicit %2
+...
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-smax.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-smax.mir
index db11855d2967b..45714fd99d7bd 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-smax.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-smax.mir
@@ -4,6 +4,7 @@
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx900 -run-pass=legalizer %s -o - | FileCheck -check-prefix=GFX9 %s
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1010 -run-pass=legalizer %s -o - | FileCheck -check-prefix=GFX9 %s
# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1100 -mattr=-real-true16 -run-pass=legalizer %s -o - | FileCheck -check-prefixes=GFX9 %s
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=gfx1250 -mattr=-real-true16 -run-pass=legalizer %s -o - | FileCheck -check-prefixes=GFX1250 %s
---
name: test_smax_s32
@@ -34,6 +35,14 @@ body: |
; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[COPY]], [[COPY1]]
; GFX9-NEXT: $vgpr0 = COPY [[SMAX]](s32)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s32
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[COPY]], [[COPY1]]
+ ; GFX1250-NEXT: $vgpr0 = COPY [[SMAX]](s32)
%0:_(s32) = COPY $vgpr0
%1:_(s32) = COPY $vgpr1
%2:_(s32) = G_SMAX %0, %1
@@ -72,6 +81,14 @@ body: |
; GFX9-NEXT: [[ICMP:%[0-9]+]]:_(s1) = G_ICMP intpred(sgt), [[COPY]](s64), [[COPY1]]
; GFX9-NEXT: [[SELECT:%[0-9]+]]:_(s64) = G_SELECT [[ICMP]](s1), [[COPY]], [[COPY1]]
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[SELECT]](s64)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s64
+ ; GFX1250: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s64) = COPY $vgpr0_vgpr1
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s64) = COPY $vgpr2_vgpr3
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s64) = G_SMAX [[COPY]], [[COPY1]]
+ ; GFX1250-NEXT: $vgpr0_vgpr1 = COPY [[SMAX]](s64)
%0:_(s64) = COPY $vgpr0_vgpr1
%1:_(s64) = COPY $vgpr2_vgpr3
%2:_(s64) = G_SMAX %0, %1
@@ -115,6 +132,17 @@ body: |
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(s16) = G_SMAX [[TRUNC]], [[TRUNC1]]
; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX]](s16)
; GFX9-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s16
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+ ; GFX1250-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[COPY]](s32)
+ ; GFX1250-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[COPY1]](s32)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s16) = G_SMAX [[TRUNC]], [[TRUNC1]]
+ ; GFX1250-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX]](s16)
+ ; GFX1250-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
%0:_(s32) = COPY $vgpr0
%1:_(s32) = COPY $vgpr1
%2:_(s16) = G_TRUNC %0
@@ -165,6 +193,19 @@ body: |
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(s16) = G_SMAX [[TRUNC]], [[TRUNC1]]
; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX]](s16)
; GFX9-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s8
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+ ; GFX1250-NEXT: [[SEXT_INREG:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY]], 8
+ ; GFX1250-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[SEXT_INREG]](s32)
+ ; GFX1250-NEXT: [[SEXT_INREG1:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY1]], 8
+ ; GFX1250-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[SEXT_INREG1]](s32)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s16) = G_SMAX [[TRUNC]], [[TRUNC1]]
+ ; GFX1250-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX]](s16)
+ ; GFX1250-NEXT: $vgpr0 = COPY [[ANYEXT]](s32)
%0:_(s32) = COPY $vgpr0
%1:_(s32) = COPY $vgpr1
%2:_(s8) = G_TRUNC %0
@@ -209,6 +250,16 @@ body: |
; GFX9-NEXT: [[SEXT_INREG1:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY1]], 17
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[SEXT_INREG]], [[SEXT_INREG1]]
; GFX9-NEXT: $vgpr0 = COPY [[SMAX]](s32)
+ ;
+ ; GFX1250-LABEL: name: test_smax_s17
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(s32) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(s32) = COPY $vgpr1
+ ; GFX1250-NEXT: [[SEXT_INREG:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY]], 17
+ ; GFX1250-NEXT: [[SEXT_INREG1:%[0-9]+]]:_(s32) = G_SEXT_INREG [[COPY1]], 17
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[SEXT_INREG]], [[SEXT_INREG1]]
+ ; GFX1250-NEXT: $vgpr0 = COPY [[SMAX]](s32)
%0:_(s32) = COPY $vgpr0
%1:_(s32) = COPY $vgpr1
%2:_(s17) = G_TRUNC %0
@@ -259,6 +310,18 @@ body: |
; GFX9-NEXT: [[SMAX1:%[0-9]+]]:_(s32) = G_SMAX [[UV1]], [[UV3]]
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[SMAX]](s32), [[SMAX1]](s32)
; GFX9-NEXT: $vgpr0_vgpr1 = COPY [[BUILD_VECTOR]](<2 x s32>)
+ ;
+ ; GFX1250-LABEL: name: test_smax_v2s32
+ ; GFX1250: liveins: $vgpr0_vgpr1, $vgpr2_vgpr3
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr0_vgpr1
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(<2 x s32>) = COPY $vgpr2_vgpr3
+ ; GFX1250-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<2 x s32>)
+ ; GFX1250-NEXT: [[UV2:%[0-9]+]]:_(s32), [[UV3:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<2 x s32>)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[UV]], [[UV2]]
+ ; GFX1250-NEXT: [[SMAX1:%[0-9]+]]:_(s32) = G_SMAX [[UV1]], [[UV3]]
+ ; GFX1250-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<2 x s32>) = G_BUILD_VECTOR [[SMAX]](s32), [[SMAX1]](s32)
+ ; GFX1250-NEXT: $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_SMAX %0, %1
@@ -309,6 +372,19 @@ body: |
; GFX9-NEXT: [[SMAX2:%[0-9]+]]:_(s32) = G_SMAX [[UV2]], [[UV5]]
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[SMAX]](s32), [[SMAX1]](s32), [[SMAX2]](s32)
; GFX9-NEXT: $vgpr0_vgpr1_vgpr2 = COPY [[BUILD_VECTOR]](<3 x s32>)
+ ;
+ ; GFX1250-LABEL: name: test_smax_v3s32
+ ; GFX1250: liveins: $vgpr0_vgpr1_vgpr2, $vgpr3_vgpr4_vgpr5
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr0_vgpr1_vgpr2
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(<3 x s32>) = COPY $vgpr3_vgpr4_vgpr5
+ ; GFX1250-NEXT: [[UV:%[0-9]+]]:_(s32), [[UV1:%[0-9]+]]:_(s32), [[UV2:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY]](<3 x s32>)
+ ; GFX1250-NEXT: [[UV3:%[0-9]+]]:_(s32), [[UV4:%[0-9]+]]:_(s32), [[UV5:%[0-9]+]]:_(s32) = G_UNMERGE_VALUES [[COPY1]](<3 x s32>)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(s32) = G_SMAX [[UV]], [[UV3]]
+ ; GFX1250-NEXT: [[SMAX1:%[0-9]+]]:_(s32) = G_SMAX [[UV1]], [[UV4]]
+ ; GFX1250-NEXT: [[SMAX2:%[0-9]+]]:_(s32) = G_SMAX [[UV2]], [[UV5]]
+ ; GFX1250-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[SMAX]](s32), [[SMAX1]](s32), [[SMAX2]](s32)
+ ; GFX1250-NEXT: $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_SMAX %0, %1
@@ -375,6 +451,14 @@ body: |
; GFX9-NEXT: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
; GFX9-NEXT: [[SMAX:%[0-9]+]]:_(<2 x s16>) = G_SMAX [[COPY]], [[COPY1]]
; GFX9-NEXT: $vgpr0 = COPY [[SMAX]](<2 x s16>)
+ ;
+ ; GFX1250-LABEL: name: test_smax_v2s16
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[COPY:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr0
+ ; GFX1250-NEXT: [[COPY1:%[0-9]+]]:_(<2 x s16>) = COPY $vgpr1
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(<2 x s16>) = G_SMAX [[COPY]], [[COPY1]]
+ ; GFX1250-NEXT: $vgpr0 = COPY [[SMAX]](<2 x s16>)
%0:_(<2 x s16>) = COPY $vgpr0
%1:_(<2 x s16>) = COPY $vgpr1
%2:_(<2 x s16>) = G_SMAX %0, %1
@@ -461,6 +545,26 @@ body: |
; GFX9-NEXT: [[ANYEXT:%[0-9]+]]:_(s32) = G_ANYEXT [[SMAX1]](s16)
; GFX9-NEXT: [[BUILD_VECTOR:%[0-9]+]]:_(<3 x s32>) = G_BUILD_VECTOR [[BITCAST2]](s32), [[LSHR]](s32), [[ANYEXT]](s32)
; GFX9-NEXT: S_NOP 0, implicit [[BUILD_VECTOR]](<3 x s32>)
+ ;
+ ; GFX1250-LABEL: name: test_smax_v3s16
+ ; GFX1250: liveins: $vgpr0, $vgpr1
+ ; GFX1250-NEXT: {{ $}}
+ ; GFX1250-NEXT: [[DEF:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
+ ; GFX1250-NEXT: [[UV:%[0-9]+]]:_(<2 x s16>), [[UV1:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[DEF]](<4 x s16>)
+ ; GFX1250-NEXT: [[BITCAST:%[0-9]+]]:_(s32) = G_BITCAST [[UV1]](<2 x s16>)
+ ; GFX1250-NEXT: [[TRUNC:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST]](s32)
+ ; GFX1250-NEXT: [[C:%[0-9]+]]:_(s32) = G_CONSTANT i32 16
+ ; GFX1250-NEXT: [[DEF1:%[0-9]+]]:_(<4 x s16>) = G_IMPLICIT_DEF
+ ; GFX1250-NEXT: [[UV2:%[0-9]+]]:_(<2 x s16>), [[UV3:%[0-9]+]]:_(<2 x s16>) = G_UNMERGE_VALUES [[DEF1]](<4 x s16>)
+ ; GFX1250-NEXT: [[BITCAST1:%[0-9]+]]:_(s32) = G_BITCAST [[UV3]](<2 x s16>)
+ ; GFX1250-NEXT: [[TRUNC1:%[0-9]+]]:_(s16) = G_TRUNC [[BITCAST1]](s32)
+ ; GFX1250-NEXT: [[SMAX:%[0-9]+]]:_(<2 x s16>...
[truncated]
|
No description provided.