[AMDGPU] Calculate number of min/max SGPRs/VGPRs for WavesPerEU instead of using...

Differential Revision: https://reviews.llvm.org/D29741

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@294627 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/AMDGPU/AMDGPUSubtarget.cpp b/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
index 8dfd89aa759d33709dd8976e654de36828198277..a924f6920939251786987d3881ef40a0ee7cd20d 100644 (file)
--- a/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
+++ b/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
@@ -190,7 +190,7 @@ std::pair<unsigned, unsigned> AMDGPUSubtarget::getFlatWorkGroupSizes(
 std::pair<unsigned, unsigned> AMDGPUSubtarget::getWavesPerEU(
   const Function &F) const {
   // Default minimum/maximum number of waves per execution unit.
-  std::pair<unsigned, unsigned> Default(1, 0);
+  std::pair<unsigned, unsigned> Default(1, getMaxWavesPerEU());
 
   // Default/requested minimum/maximum flat work group sizes.
   std::pair<unsigned, unsigned> FlatWorkGroupSizes = getFlatWorkGroupSizes(F);

diff --git a/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp b/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
index 7e454609bf73abdbd799d6c2d2e40e8efd9cf124..ec13245c0eb364ac5a49827ec1045b10f5ab9509 100644 (file)
--- a/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
+++ b/lib/Target/AMDGPU/Utils/AMDGPUBaseInfo.cpp
@@ -209,52 +209,27 @@ unsigned getAddressableNumSGPRs(const FeatureBitset &Features) {
 }
 
 unsigned getMinNumSGPRs(const FeatureBitset &Features, unsigned WavesPerEU) {
-  IsaVersion Version = getIsaVersion(Features);
-  if (Version.Major >= 8) {
-    switch (WavesPerEU) {
-      case 0:  return 0;
-      case 10: return 0;
-      case 9:  return 0;
-      case 8:  return 81;
-      default: return 97;
-    }
-  } else {
-    switch (WavesPerEU) {
-      case 0:  return 0;
-      case 10: return 0;
-      case 9:  return 49;
-      case 8:  return 57;
-      case 7:  return 65;
-      case 6:  return 73;
-      case 5:  return 81;
-      default: return 97;
-    }
-  }
+  assert(WavesPerEU != 0);
+
+  if (WavesPerEU >= getMaxWavesPerEU(Features))
+    return 0;
+  unsigned MinNumSGPRs =
+      alignDown(getTotalNumSGPRs(Features) / (WavesPerEU + 1),
+                getSGPRAllocGranule(Features)) + 1;
+  return std::min(MinNumSGPRs, getAddressableNumSGPRs(Features));
 }
 
 unsigned getMaxNumSGPRs(const FeatureBitset &Features, unsigned WavesPerEU,
                         bool Addressable) {
+  assert(WavesPerEU != 0);
+
   IsaVersion Version = getIsaVersion(Features);
-  if (Version.Major >= 8) {
-    switch (WavesPerEU) {
-      case 0:  return 80;
-      case 10: return 80;
-      case 9:  return 80;
-      case 8:  return 96;
-      default: return Addressable ? getAddressableNumSGPRs(Features) : 112;
-    }
-  } else {
-    switch (WavesPerEU) {
-      case 0:  return 48;
-      case 10: return 48;
-      case 9:  return 56;
-      case 8:  return 64;
-      case 7:  return 72;
-      case 6:  return 80;
-      case 5:  return 96;
-      default: return getAddressableNumSGPRs(Features);
-    }
-  }
+  unsigned MaxNumSGPRs = alignDown(getTotalNumSGPRs(Features) / WavesPerEU,
+                                   getSGPRAllocGranule(Features));
+  unsigned AddressableNumSGPRs = getAddressableNumSGPRs(Features);
+  if (Version.Major >= 8 && !Addressable)
+    AddressableNumSGPRs = 112;
+  return std::min(MaxNumSGPRs, AddressableNumSGPRs);
 }
 
 unsigned getVGPRAllocGranule(const FeatureBitset &Features) {
@@ -274,35 +249,23 @@ unsigned getAddressableNumVGPRs(const FeatureBitset &Features) {
 }
 
 unsigned getMinNumVGPRs(const FeatureBitset &Features, unsigned WavesPerEU) {
-  switch (WavesPerEU) {
-    case 0:  return 0;
-    case 10: return 0;
-    case 9:  return 25;
-    case 8:  return 29;
-    case 7:  return 33;
-    case 6:  return 37;
-    case 5:  return 41;
-    case 4:  return 49;
-    case 3:  return 65;
-    case 2:  return 85;
-    default: return 129;
-  }
+  assert(WavesPerEU != 0);
+
+  if (WavesPerEU >= getMaxWavesPerEU(Features))
+    return 0;
+  unsigned MinNumVGPRs =
+      alignDown(getTotalNumVGPRs(Features) / (WavesPerEU + 1),
+                getVGPRAllocGranule(Features)) + 1;
+  return std::min(MinNumVGPRs, getAddressableNumVGPRs(Features));
 }
 
 unsigned getMaxNumVGPRs(const FeatureBitset &Features, unsigned WavesPerEU) {
-  switch (WavesPerEU) {
-    case 0:  return 24;
-    case 10: return 24;
-    case 9:  return 28;
-    case 8:  return 32;
-    case 7:  return 36;
-    case 6:  return 40;
-    case 5:  return 48;
-    case 4:  return 64;
-    case 3:  return 84;
-    case 2:  return 128;
-    default: return getTotalNumVGPRs(Features);
-  }
+  assert(WavesPerEU != 0);
+
+  unsigned MaxNumVGPRs = alignDown(getTotalNumVGPRs(Features) / WavesPerEU,
+                                   getVGPRAllocGranule(Features));
+  unsigned AddressableNumVGPRs = getAddressableNumVGPRs(Features);
+  return std::min(MaxNumVGPRs, AddressableNumVGPRs);
 }
 
 } // namespace IsaInfo

diff --git a/test/CodeGen/AMDGPU/attr-amdgpu-waves-per-eu.ll b/test/CodeGen/AMDGPU/attr-amdgpu-waves-per-eu.ll
index 4f4efccc2260a4f5c189b50a0ec640f080dc0961..c1c334283e600649a4969b9ab78adcbaa7d2ef99 100644 (file)
--- a/test/CodeGen/AMDGPU/attr-amdgpu-waves-per-eu.ll
+++ b/test/CodeGen/AMDGPU/attr-amdgpu-waves-per-eu.ll
@@ -4,7 +4,7 @@
 ; CHECK-LABEL: {{^}}empty_exactly_1:
 ; CHECK: SGPRBlocks: 12
 ; CHECK: VGPRBlocks: 32
-; CHECK: NumSGPRsForWavesPerEU: 97
+; CHECK: NumSGPRsForWavesPerEU: 102
 ; CHECK: NumVGPRsForWavesPerEU: 129
 define void @empty_exactly_1() #0 {
 entry:
@@ -16,7 +16,7 @@ attributes #0 = {"amdgpu-waves-per-eu"="1,1"}
 ; CHECK-LABEL: {{^}}empty_exactly_5:
 ; CHECK: SGPRBlocks: 12
 ; CHECK: VGPRBlocks: 10
-; CHECK: NumSGPRsForWavesPerEU: 97
+; CHECK: NumSGPRsForWavesPerEU: 102
 ; CHECK: NumVGPRsForWavesPerEU: 41
 define void @empty_exactly_5() #1 {
 entry:
@@ -78,7 +78,7 @@ attributes #5 = {"amdgpu-waves-per-eu"="10"}
 ; CHECK-LABEL: {{^}}empty_at_most_5:
 ; CHECK: SGPRBlocks: 12
 ; CHECK: VGPRBlocks: 10
-; CHECK: NumSGPRsForWavesPerEU: 97
+; CHECK: NumSGPRsForWavesPerEU: 102
 ; CHECK: NumVGPRsForWavesPerEU: 41
 define void @empty_at_most_5() #6 {
 entry: