[llvm] [AMDGPU][InsertWaitCnts] Add test for global_wb/inv/wbinv tracking (PR #135339)
Pierre van Houtryve via llvm-commits
llvm-commits at lists.llvm.org
Tue Apr 22 05:44:49 PDT 2025
https://github.com/Pierre-vh updated https://github.com/llvm/llvm-project/pull/135339
>From b1422e75902f7186d6265158797abc402d03908f Mon Sep 17 00:00:00 2001
From: pvanhout <pierre.vanhoutryve at amd.com>
Date: Fri, 11 Apr 2025 11:45:24 +0200
Subject: [PATCH] [AMDGPU][InsertWaitCnts] Add test for global_wb/inv/wbinv
tracking
---
.../AMDGPU/insert-waitcnts-gfx12-wbinv.mir | 129 ++++++++++++++++++
1 file changed, 129 insertions(+)
create mode 100644 llvm/test/CodeGen/AMDGPU/insert-waitcnts-gfx12-wbinv.mir
diff --git a/llvm/test/CodeGen/AMDGPU/insert-waitcnts-gfx12-wbinv.mir b/llvm/test/CodeGen/AMDGPU/insert-waitcnts-gfx12-wbinv.mir
new file mode 100644
index 0000000000000..6758a43aedd3c
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/insert-waitcnts-gfx12-wbinv.mir
@@ -0,0 +1,129 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 5
+# RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -run-pass si-insert-waitcnts -o - %s | FileCheck %s
+
+
+# Check storecnt is not eliminated after global_wb/wbinv or
+# loadcnt after inv.
+---
+
+name: wb_storecnt
+body: |
+ ; CHECK-LABEL: name: wb_storecnt
+ ; CHECK: bb.0:
+ ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: S_WAIT_LOADCNT_DSCNT 0
+ ; CHECK-NEXT: S_WAIT_EXPCNT 0
+ ; CHECK-NEXT: S_WAIT_SAMPLECNT 0
+ ; CHECK-NEXT: S_WAIT_BVHCNT 0
+ ; CHECK-NEXT: S_WAIT_KMCNT 0
+ ; CHECK-NEXT: S_WAIT_STORECNT 0
+ ; CHECK-NEXT: S_CMP_EQ_U32 renamable $sgpr1, 0, implicit-def $scc
+ ; CHECK-NEXT: S_CBRANCH_SCC1 %bb.2, implicit killed $scc
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.1:
+ ; CHECK-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: GLOBAL_WB 16, implicit $exec
+ ; CHECK-NEXT: GLOBAL_STORE_DWORD_SADDR renamable $vgpr0, renamable $vgpr1, renamable $sgpr2_sgpr3, 0, 16, implicit $exec
+ ; CHECK-NEXT: S_CBRANCH_SCC0 %bb.1, implicit killed $scc
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.2:
+ ; CHECK-NEXT: S_ENDPGM 0
+ bb.0:
+ S_WAIT_LOADCNT_soft 0
+ S_WAIT_STORECNT_soft 0
+ S_WAIT_DSCNT_soft 0
+ S_CMP_EQ_U32 renamable $sgpr1, 0, implicit-def $scc
+ S_CBRANCH_SCC1 %bb.2, implicit killed $scc
+ bb.1:
+ GLOBAL_WB 16, implicit $exec
+ S_WAIT_LOADCNT_soft 0
+ S_WAIT_STORECNT_soft 0
+ GLOBAL_STORE_DWORD_SADDR renamable $vgpr0, renamable $vgpr1, renamable $sgpr2_sgpr3, 0, 16, implicit $exec
+ S_CBRANCH_SCC0 %bb.1, implicit killed $scc
+ bb.2:
+ S_ENDPGM 0
+...
+---
+
+name: wbinv_storecnt
+body: |
+ ; CHECK-LABEL: name: wbinv_storecnt
+ ; CHECK: bb.0:
+ ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: S_WAIT_LOADCNT_DSCNT 0
+ ; CHECK-NEXT: S_WAIT_EXPCNT 0
+ ; CHECK-NEXT: S_WAIT_SAMPLECNT 0
+ ; CHECK-NEXT: S_WAIT_BVHCNT 0
+ ; CHECK-NEXT: S_WAIT_KMCNT 0
+ ; CHECK-NEXT: S_WAIT_STORECNT 0
+ ; CHECK-NEXT: S_CMP_EQ_U32 renamable $sgpr1, 0, implicit-def $scc
+ ; CHECK-NEXT: S_CBRANCH_SCC1 %bb.2, implicit killed $scc
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.1:
+ ; CHECK-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: GLOBAL_WBINV 16, implicit $exec
+ ; CHECK-NEXT: GLOBAL_STORE_DWORD_SADDR renamable $vgpr0, renamable $vgpr1, renamable $sgpr2_sgpr3, 0, 16, implicit $exec
+ ; CHECK-NEXT: S_CBRANCH_SCC0 %bb.1, implicit killed $scc
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.2:
+ ; CHECK-NEXT: S_ENDPGM 0
+ bb.0:
+ S_WAIT_LOADCNT_soft 0
+ S_WAIT_STORECNT_soft 0
+ S_WAIT_DSCNT_soft 0
+ S_CMP_EQ_U32 renamable $sgpr1, 0, implicit-def $scc
+ S_CBRANCH_SCC1 %bb.2, implicit killed $scc
+ bb.1:
+ GLOBAL_WBINV 16, implicit $exec
+ S_WAIT_LOADCNT_soft 0
+ S_WAIT_STORECNT_soft 0
+ GLOBAL_STORE_DWORD_SADDR renamable $vgpr0, renamable $vgpr1, renamable $sgpr2_sgpr3, 0, 16, implicit $exec
+ S_CBRANCH_SCC0 %bb.1, implicit killed $scc
+ bb.2:
+ S_ENDPGM 0
+...
+---
+
+name: inv_loadcnt
+body: |
+ ; CHECK-LABEL: name: inv_loadcnt
+ ; CHECK: bb.0:
+ ; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.1(0x40000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: S_WAIT_LOADCNT_DSCNT 0
+ ; CHECK-NEXT: S_WAIT_EXPCNT 0
+ ; CHECK-NEXT: S_WAIT_SAMPLECNT 0
+ ; CHECK-NEXT: S_WAIT_BVHCNT 0
+ ; CHECK-NEXT: S_WAIT_KMCNT 0
+ ; CHECK-NEXT: S_WAIT_STORECNT 0
+ ; CHECK-NEXT: S_CMP_EQ_U32 renamable $sgpr1, 0, implicit-def $scc
+ ; CHECK-NEXT: S_CBRANCH_SCC1 %bb.2, implicit killed $scc
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.1:
+ ; CHECK-NEXT: successors: %bb.1(0x40000000), %bb.2(0x40000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: GLOBAL_INV 16, implicit $exec
+ ; CHECK-NEXT: GLOBAL_STORE_DWORD_SADDR renamable $vgpr0, renamable $vgpr1, renamable $sgpr2_sgpr3, 0, 16, implicit $exec
+ ; CHECK-NEXT: S_CBRANCH_SCC0 %bb.1, implicit killed $scc
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.2:
+ ; CHECK-NEXT: S_ENDPGM 0
+ bb.0:
+ S_WAIT_LOADCNT_soft 0
+ S_WAIT_STORECNT_soft 0
+ S_WAIT_DSCNT_soft 0
+ S_CMP_EQ_U32 renamable $sgpr1, 0, implicit-def $scc
+ S_CBRANCH_SCC1 %bb.2, implicit killed $scc
+ bb.1:
+ GLOBAL_INV 16, implicit $exec
+ S_WAIT_LOADCNT_soft 0
+ S_WAIT_STORECNT_soft 0
+ GLOBAL_STORE_DWORD_SADDR renamable $vgpr0, renamable $vgpr1, renamable $sgpr2_sgpr3, 0, 16, implicit $exec
+ S_CBRANCH_SCC0 %bb.1, implicit killed $scc
+ bb.2:
+ S_ENDPGM 0
+...
More information about the llvm-commits
mailing list