[AMDGPU][Attributor] Make `AAAMDWavesPerEU` honor existing attribute #114438

shiltian · 2024-10-31T17:54:36Z

No description provided.

shiltian · 2024-10-31T17:54:52Z

This stack of pull requests is managed by Graphite. Learn more about stacking.

llvmbot · 2024-10-31T17:55:20Z

@llvm/pr-subscribers-backend-amdgpu

Author: Shilei Tian (shiltian)

Changes

Patch is 25.38 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/114438.diff

2 Files Affected:

(modified) llvm/lib/Target/AMDGPU/AMDGPUAttributor.cpp (+46-18)
(modified) llvm/test/CodeGen/AMDGPU/propagate-waves-per-eu.ll (+23-24)

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUAttributor.cpp b/llvm/lib/Target/AMDGPU/AMDGPUAttributor.cpp
index 205673cdcc0e23..ed7cd1f53b41e9 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUAttributor.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUAttributor.cpp
@@ -215,6 +215,19 @@ class AMDGPUInformationCache : public InformationCache {
     return ST.getWavesPerEU(F, FlatWorkGroupSize);
   }
 
+  std::optional<std::pair<unsigned, unsigned>>
+  getWavesPerEUAttr(const Function &F) {
+    Attribute Attr = F.getFnAttribute("amdgpu-waves-per-eu");
+    if (!Attr.isStringAttribute())
+      return std::nullopt;
+    auto Val = parseRangeAttribute(Attr.getValueAsString());
+    if (Val && Val->second == 0) {
+      const GCNSubtarget &ST = TM.getSubtarget<GCNSubtarget>(F);
+      Val->second = ST.getMaxWavesPerEU();
+    }
+    return Val;
+  }
+
   std::pair<unsigned, unsigned>
   getEffectiveWavesPerEU(const Function &F,
                          std::pair<unsigned, unsigned> WavesPerEU,
@@ -885,29 +898,44 @@ struct AAAMDWavesPerEU : public AAAMDSizeRangeAttribute {
   AAAMDWavesPerEU(const IRPosition &IRP, Attributor &A)
       : AAAMDSizeRangeAttribute(IRP, A, "amdgpu-waves-per-eu") {}
 
-  bool isValidState() const override {
-    return !Assumed.isEmptySet() && IntegerRangeState::isValidState();
-  }
-
   void initialize(Attributor &A) override {
     Function *F = getAssociatedFunction();
     auto &InfoCache = static_cast<AMDGPUInformationCache &>(A.getInfoCache());
 
-    if (const auto *AssumedGroupSize = A.getAAFor<AAAMDFlatWorkGroupSize>(
-            *this, IRPosition::function(*F), DepClassTy::REQUIRED);
-        AssumedGroupSize->isValidState()) {
-
-      unsigned Min, Max;
-      std::tie(Min, Max) = InfoCache.getWavesPerEU(
-          *F, {AssumedGroupSize->getAssumed().getLower().getZExtValue(),
-               AssumedGroupSize->getAssumed().getUpper().getZExtValue() - 1});
-
+    auto TakeRange = [&](std::pair<unsigned, unsigned> R) {
+      auto [Min, Max] = R;
       ConstantRange Range(APInt(32, Min), APInt(32, Max + 1));
-      intersectKnown(Range);
+      IntegerRangeState RangeState(Range);
+      clampStateAndIndicateChange(this->getState(), RangeState);
+      indicateOptimisticFixpoint();
+    };
+
+    // If the attribute exists, simple honor it.
+    if (auto Attr = InfoCache.getWavesPerEUAttr(*F)) {
+      TakeRange(*Attr);
+      return;
     }
 
-    if (AMDGPU::isEntryFunctionCC(F->getCallingConv()))
-      indicatePessimisticFixpoint();
+    // It's getting trickier here, different from AAAMDFlatWorkGroupSize. Since
+    // the calculation of waves per EU involves flat work group size, we can't
+    // simply use an assumed flat work group size as a start point, because the
+    // update of flat work group size is in an inverse direction of waves per
+    // EU. However, we can still do something if it is an entry function. Since
+    // an entry function is a terminal node, and flat work group size either
+    // from attribute or default will be used anyway, we can take that value and
+    // calculate the waves per EU based on it. This result can't be updated by
+    // no means, but that could still allow us to propagate it.
+    if (AMDGPU::isEntryFunctionCC(F->getCallingConv())) {
+      std::pair<unsigned, unsigned> MaxWavesPerEURange{
+          1U, InfoCache.getMaxWavesPerEU(*F)};
+      std::pair<unsigned, unsigned> FlatWorkGroupSize;
+      if (auto Attr = InfoCache.getFlatWorkGroupSizeAttr(*F))
+        FlatWorkGroupSize = *Attr;
+      else
+        FlatWorkGroupSize = InfoCache.getDefaultFlatWorkGroupSize(*F);
+      TakeRange(InfoCache.getEffectiveWavesPerEU(*F, MaxWavesPerEURange,
+                                                 FlatWorkGroupSize));
+    }
   }
 
   ChangeStatus updateImpl(Attributor &A) override {
@@ -956,8 +984,8 @@ struct AAAMDWavesPerEU : public AAAMDSizeRangeAttribute {
   ChangeStatus manifest(Attributor &A) override {
     Function *F = getAssociatedFunction();
     auto &InfoCache = static_cast<AMDGPUInformationCache &>(A.getInfoCache());
-    unsigned Max = InfoCache.getMaxWavesPerEU(*F);
-    return emitAttributeIfNotDefault(A, 1, Max);
+    return emitAttributeIfNotDefaultAfterClamp(
+        A, {1, InfoCache.getMaxWavesPerEU(*F)});
   }
 
   /// See AbstractAttribute::getName()
diff --git a/llvm/test/CodeGen/AMDGPU/propagate-waves-per-eu.ll b/llvm/test/CodeGen/AMDGPU/propagate-waves-per-eu.ll
index 1272bf655e309d..e28bccfb99343b 100644
--- a/llvm/test/CodeGen/AMDGPU/propagate-waves-per-eu.ll
+++ b/llvm/test/CodeGen/AMDGPU/propagate-waves-per-eu.ll
@@ -117,7 +117,7 @@ define amdgpu_kernel void @kernel_2_9() #6 {
 
 define amdgpu_kernel void @kernel_9_9() #7 {
 ; CHECK-LABEL: define amdgpu_kernel void @kernel_9_9
-; CHECK-SAME: () #[[ATTR6]] {
+; CHECK-SAME: () #[[ATTR8:[0-9]+]] {
 ; CHECK-NEXT:    call void @default_to_2_9()
 ; CHECK-NEXT:    call void @flat_group_9_10()
 ; CHECK-NEXT:    ret void
@@ -140,7 +140,7 @@ define internal void @default_to_1_8_b() {
 ; this should probably be illegal.
 define amdgpu_kernel void @kernel_2_8() #4 {
 ; CHECK-LABEL: define amdgpu_kernel void @kernel_2_8
-; CHECK-SAME: () #[[ATTR8:[0-9]+]] {
+; CHECK-SAME: () #[[ATTR5]] {
 ; CHECK-NEXT:    call void @default_to_1_8_a()
 ; CHECK-NEXT:    call void @default_to_1_8_b()
 ; CHECK-NEXT:    ret void
@@ -153,7 +153,7 @@ define amdgpu_kernel void @kernel_2_8() #4 {
 ; 1,2 -> 2,2
 define internal void @merge_cycle_0() #1 {
 ; CHECK-LABEL: define internal void @merge_cycle_0
-; CHECK-SAME: () #[[ATTR5]] {
+; CHECK-SAME: () #[[ATTR2]] {
 ; CHECK-NEXT:    call void @merge_cycle_1()
 ; CHECK-NEXT:    ret void
 ;
@@ -165,7 +165,7 @@ define internal void @merge_cycle_0() #1 {
 ; 2,8 -> 2,8
 define internal void @merge_cycle_1() #4 {
 ; CHECK-LABEL: define internal void @merge_cycle_1
-; CHECK-SAME: () #[[ATTR8]] {
+; CHECK-SAME: () #[[ATTR5]] {
 ; CHECK-NEXT:    call void @merge_cycle_0()
 ; CHECK-NEXT:    ret void
 ;
@@ -235,7 +235,7 @@ define internal void @called_from_invalid_bounds_1() {
 ; Invalid range for amdgpu-waves-per-eu
 define amdgpu_kernel void @kernel_invalid_bounds_0_8() #9 {
 ; CHECK-LABEL: define amdgpu_kernel void @kernel_invalid_bounds_0_8
-; CHECK-SAME: () #[[ATTR11:[0-9]+]] {
+; CHECK-SAME: () #[[ATTR1]] {
 ; CHECK-NEXT:    call void @called_from_invalid_bounds_0()
 ; CHECK-NEXT:    ret void
 ;
@@ -246,7 +246,7 @@ define amdgpu_kernel void @kernel_invalid_bounds_0_8() #9 {
 ; Invalid range for amdgpu-waves-per-eu
 define amdgpu_kernel void @kernel_invalid_bounds_1_123() #10 {
 ; CHECK-LABEL: define amdgpu_kernel void @kernel_invalid_bounds_1_123
-; CHECK-SAME: () #[[ATTR12:[0-9]+]] {
+; CHECK-SAME: () #[[ATTR11:[0-9]+]] {
 ; CHECK-NEXT:    call void @called_from_invalid_bounds_1()
 ; CHECK-NEXT:    ret void
 ;
@@ -260,7 +260,7 @@ define amdgpu_kernel void @kernel_invalid_bounds_1_123() #10 {
 ; -> 2,10
 define void @larger_group_size_implies_lower_minimum() #11 {
 ; CHECK-LABEL: define void @larger_group_size_implies_lower_minimum
-; CHECK-SAME: () #[[ATTR13:[0-9]+]] {
+; CHECK-SAME: () #[[ATTR12:[0-9]+]] {
 ; CHECK-NEXT:    ret void
 ;
   ret void
@@ -268,7 +268,7 @@ define void @larger_group_size_implies_lower_minimum() #11 {
 
 define amdgpu_kernel void @kernel_3_6() #12 {
 ; CHECK-LABEL: define amdgpu_kernel void @kernel_3_6
-; CHECK-SAME: () #[[ATTR14:[0-9]+]] {
+; CHECK-SAME: () #[[ATTR13:[0-9]+]] {
 ; CHECK-NEXT:    call void @larger_group_size_implies_lower_minimum()
 ; CHECK-NEXT:    ret void
 ;
@@ -279,7 +279,7 @@ define amdgpu_kernel void @kernel_3_6() #12 {
 ; 3,6 -> 6,9
 define internal void @refine_upper_func_3_6() #13 {
 ; CHECK-LABEL: define internal void @refine_upper_func_3_6
-; CHECK-SAME: () #[[ATTR15:[0-9]+]] {
+; CHECK-SAME: () #[[ATTR14:[0-9]+]] {
 ; CHECK-NEXT:    ret void
 ;
   ret void
@@ -288,7 +288,7 @@ define internal void @refine_upper_func_3_6() #13 {
 ; 4,8 -> 6,8
 define internal void @refine_lower_func_4_8() #14 {
 ; CHECK-LABEL: define internal void @refine_lower_func_4_8
-; CHECK-SAME: () #[[ATTR16:[0-9]+]] {
+; CHECK-SAME: () #[[ATTR15:[0-9]+]] {
 ; CHECK-NEXT:    call void @refine_upper_func_3_6()
 ; CHECK-NEXT:    ret void
 ;
@@ -298,7 +298,7 @@ define internal void @refine_lower_func_4_8() #14 {
 
 define amdgpu_kernel void @kernel_foo_6_8() #15 {
 ; CHECK-LABEL: define amdgpu_kernel void @kernel_foo_6_8
-; CHECK-SAME: () #[[ATTR16]] {
+; CHECK-SAME: () #[[ATTR16:[0-9]+]] {
 ; CHECK-NEXT:    call void @refine_upper_func_3_6()
 ; CHECK-NEXT:    call void @refine_lower_func_4_8()
 ; CHECK-NEXT:    call void @func_9_10_a()
@@ -340,7 +340,7 @@ define internal void @func_9_10_a() #18 {
 ; 9,10 -> 9,9
 define internal void @func_9_10_b() #18 {
 ; CHECK-LABEL: define internal void @func_9_10_b
-; CHECK-SAME: () #[[ATTR20:[0-9]+]] {
+; CHECK-SAME: () #[[ATTR19]] {
 ; CHECK-NEXT:    ret void
 ;
   ret void
@@ -348,7 +348,7 @@ define internal void @func_9_10_b() #18 {
 
 define amdgpu_kernel void @kernel_bar_8_9() #19 {
 ; CHECK-LABEL: define amdgpu_kernel void @kernel_bar_8_9
-; CHECK-SAME: () #[[ATTR21:[0-9]+]] {
+; CHECK-SAME: () #[[ATTR20:[0-9]+]] {
 ; CHECK-NEXT:    call void @refine_upper_func_3_6()
 ; CHECK-NEXT:    call void @func_5_5()
 ; CHECK-NEXT:    call void @func_9_10_b()
@@ -404,21 +404,20 @@ attributes #19 = { "amdgpu-waves-per-eu"="8,9" }
 ; CHECK: attributes #[[ATTR2]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="1,2" "uniform-work-group-size"="false" }
 ; CHECK: attributes #[[ATTR3]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="1,4" "uniform-work-group-size"="false" }
 ; CHECK: attributes #[[ATTR4]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="1,1" "uniform-work-group-size"="false" }
-; CHECK: attributes #[[ATTR5]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="2,2" "uniform-work-group-size"="false" }
-; CHECK: attributes #[[ATTR6]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="9,9" "uniform-work-group-size"="false" }
+; CHECK: attributes #[[ATTR5]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="2,8" "uniform-work-group-size"="false" }
+; CHECK: attributes #[[ATTR6]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="9,10" "uniform-work-group-size"="false" }
 ; CHECK: attributes #[[ATTR7]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="2,9" "uniform-work-group-size"="false" }
-; CHECK: attributes #[[ATTR8]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="2,8" "uniform-work-group-size"="false" }
+; CHECK: attributes #[[ATTR8]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="9,9" "uniform-work-group-size"="false" }
 ; CHECK: attributes #[[ATTR9]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="3,8" "uniform-work-group-size"="false" }
 ; CHECK: attributes #[[ATTR10]] = { "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" }
-; CHECK: attributes #[[ATTR11]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="0,8" "uniform-work-group-size"="false" }
-; CHECK: attributes #[[ATTR12]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="1,123" "uniform-work-group-size"="false" }
-; CHECK: attributes #[[ATTR13]] = { "amdgpu-flat-work-group-size"="1,512" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="2,10" "uniform-work-group-size"="false" }
-; CHECK: attributes #[[ATTR14]] = { "amdgpu-flat-work-group-size"="1,512" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="3,6" "uniform-work-group-size"="false" }
-; CHECK: attributes #[[ATTR15]] = { "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="6,9" "uniform-work-group-size"="false" }
+; CHECK: attributes #[[ATTR11]] = { "amdgpu-flat-work-group-size"="1,64" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "amdgpu-waves-per-eu"="1,123" "uniform-work-group-size"="false" }
+; CHECK: attributes #[[ATTR12]] = { "amdgpu-flat-work-group-size"="1,512" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgroup-id-z" "amdgpu-no-workitem-id-x" "amdgpu-no-workitem-id-y" "amdgpu-no-workitem-id-z" "uniform-work-group-size"="false" }
+; CHECK: attributes #[[ATTR13]] = { "amdgpu-flat-work-group-size"="1,512" "amdgpu-no-agpr" "amdgpu-no-completion-action" "amdgpu-no-default-queue" "amdgpu-no-dispatch-id" "amdgpu-no-dispatch-ptr" "amdgpu-no-heap-ptr" "amdgpu-no-hostcall-ptr" "amdgpu-no-implicitarg-ptr" "amdgpu-no-lds-kernel-id" "amdgpu-no-multigrid-sync-arg" "amdgpu-no-queue-ptr" "amdgpu-no-workgroup-id-x" "amdgpu-no-workgroup-id-y" "amdgpu-no-workgrou...
[truncated]

llvm/test/CodeGen/AMDGPU/propagate-waves-per-eu.ll

llvm/lib/Target/AMDGPU/AMDGPUAttributor.cpp

shiltian · 2024-11-05T17:49:37Z

bump

choikwa · 2024-11-12T17:02:41Z

LGTM

shiltian · 2024-12-11T21:38:11Z

Merge activity

Dec 11, 4:38 PM EST: A user started a stack merge that includes this pull request via Graphite.
Dec 11, 4:48 PM EST: Graphite rebased this pull request as part of a merge.
Dec 11, 4:50 PM EST: A user merged this pull request with Graphite.

…bute

shiltian mentioned this pull request Oct 31, 2024

[AMDGPU][Attributor] Make AAAMDFlatWorkGroupSize honor existing attribute #114357

Merged

llvmbot added the backend:AMDGPU label Oct 31, 2024

shiltian requested review from arsenm and choikwa October 31, 2024 17:54

shiltian commented Oct 31, 2024

View reviewed changes

llvm/test/CodeGen/AMDGPU/propagate-waves-per-eu.ll Outdated Show resolved Hide resolved

shiltian changed the title ~~[WIP][AMDGPU][Attributor] Make AAAMDWavesPerEU honor existing attribute~~ [AMDGPU][Attributor] Make AAAMDWavesPerEU honor existing attribute Oct 31, 2024

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from 687d29a to 66264a1 Compare October 31, 2024 18:05

arsenm reviewed Oct 31, 2024

View reviewed changes

llvm/lib/Target/AMDGPU/AMDGPUAttributor.cpp Outdated Show resolved Hide resolved

shiltian force-pushed the users/shiltian/honor-attributes branch from f1aea50 to 704650b Compare October 31, 2024 20:22

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from 66264a1 to b7f1c2b Compare October 31, 2024 20:22

shiltian force-pushed the users/shiltian/honor-attributes branch from 704650b to 6056c31 Compare October 31, 2024 21:57

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch 3 times, most recently from b485662 to d0ec41b Compare October 31, 2024 22:10

shiltian force-pushed the users/shiltian/honor-attributes branch from 6056c31 to c35f519 Compare November 2, 2024 00:07

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from d0ec41b to fe19790 Compare November 2, 2024 00:07

shiltian force-pushed the users/shiltian/honor-attributes branch from c35f519 to 2803f73 Compare November 2, 2024 03:33

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from fe19790 to 7181479 Compare November 2, 2024 03:33

shiltian force-pushed the users/shiltian/honor-attributes branch from 2803f73 to 03b2bd0 Compare November 2, 2024 14:57

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from 7181479 to 1aaf29f Compare November 2, 2024 14:58

shiltian force-pushed the users/shiltian/honor-attributes branch from 03b2bd0 to fe201bc Compare November 2, 2024 15:40

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from 1aaf29f to 013d7ae Compare November 2, 2024 15:40

shiltian force-pushed the users/shiltian/honor-attributes branch from fe201bc to b6b851f Compare November 4, 2024 00:35

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from 013d7ae to 79f8883 Compare November 4, 2024 00:36

shiltian mentioned this pull request Nov 4, 2024

[AMDGPU][Attributor] Skip update and manifest if an AA is at its initial state #114726

Closed

shiltian mentioned this pull request Nov 5, 2024

AMDGPU: Propagate amdgpu-max-num-workgroups attribute #113018

Merged

choikwa approved these changes Nov 12, 2024

View reviewed changes

shiltian force-pushed the users/shiltian/honor-attributes branch from b6b851f to 43322a4 Compare December 9, 2024 16:13

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from 79f8883 to de91e17 Compare December 9, 2024 16:13

shiltian force-pushed the users/shiltian/honor-attributes branch from 43322a4 to 2d1641d Compare December 10, 2024 17:47

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from de91e17 to eb0b434 Compare December 10, 2024 17:47

shiltian force-pushed the users/shiltian/honor-attributes branch from 2d1641d to d949823 Compare December 11, 2024 16:59

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from eb0b434 to afda3da Compare December 11, 2024 16:59

shiltian force-pushed the users/shiltian/honor-attributes branch 3 times, most recently from 768a316 to a062b29 Compare December 11, 2024 21:45

Base automatically changed from users/shiltian/honor-attributes to main December 11, 2024 21:47

[WIP][AMDGPU][Attributor] Make AAAMDWavesPerEU honor existing attri…

c9dd751

…bute

shiltian force-pushed the users/shiltian/honor-attribute-wave-per-eu branch from afda3da to c9dd751 Compare December 11, 2024 21:48

shiltian merged commit f403727 into main Dec 11, 2024
5 of 7 checks passed

shiltian deleted the users/shiltian/honor-attribute-wave-per-eu branch December 11, 2024 21:50

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[AMDGPU][Attributor] Make `AAAMDWavesPerEU` honor existing attribute #114438

[AMDGPU][Attributor] Make `AAAMDWavesPerEU` honor existing attribute #114438

Uh oh!

shiltian commented Oct 31, 2024

Uh oh!

shiltian commented Oct 31, 2024 •

edited

Loading

Uh oh!

llvmbot commented Oct 31, 2024

Uh oh!

Uh oh!

Uh oh!

shiltian commented Nov 5, 2024

Uh oh!

choikwa commented Nov 12, 2024

Uh oh!

shiltian commented Dec 11, 2024 •

edited

Loading

Uh oh!

Uh oh!

Uh oh!

[AMDGPU][Attributor] Make AAAMDWavesPerEU honor existing attribute #114438

[AMDGPU][Attributor] Make AAAMDWavesPerEU honor existing attribute #114438

Uh oh!

Conversation

shiltian commented Oct 31, 2024

Uh oh!

shiltian commented Oct 31, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

llvmbot commented Oct 31, 2024

Uh oh!

Uh oh!

Uh oh!

shiltian commented Nov 5, 2024

Uh oh!

choikwa commented Nov 12, 2024

Uh oh!

shiltian commented Dec 11, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Merge activity

Uh oh!

Uh oh!

Uh oh!

[AMDGPU][Attributor] Make `AAAMDWavesPerEU` honor existing attribute #114438

[AMDGPU][Attributor] Make `AAAMDWavesPerEU` honor existing attribute #114438

shiltian commented Oct 31, 2024 •

edited

Loading

shiltian commented Dec 11, 2024 •

edited

Loading