[PATCH] D138216: [AMDGPU] Intrinsic to expose s_wait_event for export ready
David Stuttard via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Thu Nov 17 08:53:17 PST 2022
dstuttard updated this revision to Diff 476149.
dstuttard added a comment.
Updating based on review comments
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D138216/new/
https://reviews.llvm.org/D138216
Files:
llvm/include/llvm/IR/IntrinsicsAMDGPU.td
llvm/lib/Target/AMDGPU/SOPInstructions.td
llvm/test/CodeGen/AMDGPU/llvm.amdgcn.wait.event.ll
Index: llvm/test/CodeGen/AMDGPU/llvm.amdgcn.wait.event.ll
===================================================================
--- /dev/null
+++ llvm/test/CodeGen/AMDGPU/llvm.amdgcn.wait.event.ll
@@ -0,0 +1,16 @@
+; RUN: llc -march=amdgcn -verify-machineinstrs -mcpu=gfx1100 < %s | FileCheck -check-prefix=GCN %s
+; RUN: llc -global-isel -march=amdgcn -verify-machineinstrs -mcpu=gfx1100 < %s | FileCheck -check-prefix=GCN %s
+
+; GCN-LABEL: {{^}}test_wait_event:
+; GCN: s_wait_event 0x0
+
+define amdgpu_ps void @test_wait_event() #0 {
+entry:
+ call void @llvm.amdgcn.wait.event.export.ready() #1
+ ret void
+}
+
+declare void @llvm.amdgcn.wait.event.export.ready() #0
+
+attributes #0 = { nounwind }
+attributes #1 = { convergent nounwind }
Index: llvm/lib/Target/AMDGPU/SOPInstructions.td
===================================================================
--- llvm/lib/Target/AMDGPU/SOPInstructions.td
+++ llvm/lib/Target/AMDGPU/SOPInstructions.td
@@ -1388,7 +1388,9 @@
let SubtargetPredicate = isGFX11Plus in {
def S_WAIT_EVENT : SOPP_Pseudo<"s_wait_event", (ins s16imm:$simm16),
- "$simm16">;
+ "$simm16"> {
+ let hasSideEffects = 1;
+ }
def S_DELAY_ALU : SOPP_Pseudo<"s_delay_alu", (ins DELAY_FLAG:$simm16),
"$simm16">;
} // End SubtargetPredicate = isGFX11Plus
@@ -1430,6 +1432,10 @@
(S_SEXT_I32_I16 $src)
>;
+def : GCNPat <
+ (int_amdgcn_wait_event_export_ready),
+ (S_WAIT_EVENT (i16 0))
+>;
//===----------------------------------------------------------------------===//
// SOP2 Patterns
Index: llvm/include/llvm/IR/IntrinsicsAMDGPU.td
===================================================================
--- llvm/include/llvm/IR/IntrinsicsAMDGPU.td
+++ llvm/include/llvm/IR/IntrinsicsAMDGPU.td
@@ -2067,6 +2067,9 @@
def int_amdgcn_wmma_i32_16x16x16_iu8 : AMDGPUWmmaIntrinsicIU<llvm_v4i32_ty, llvm_anyint_ty>;
def int_amdgcn_wmma_i32_16x16x16_iu4 : AMDGPUWmmaIntrinsicIU<llvm_v2i32_ty, llvm_anyint_ty>;
+def int_amdgcn_wait_event_export_ready :
+ Intrinsic<[], [], [IntrNoMem, IntrHasSideEffects, IntrWillReturn]
+>;
//===----------------------------------------------------------------------===//
// Deep learning intrinsics.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: D138216.476149.patch
Type: text/x-patch
Size: 2341 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20221117/257066c1/attachment.bin>
More information about the llvm-commits
mailing list