AMDGPU: Change internal tracking of wave size

Store the log2 wave size instead of forcing division and log2
operations when querying either.

3 files changed, 19 insertions, 19 deletions

diff --git a/llvm/lib/Target/AMDGPU/AMDGPUFeatures.td b/llvm/lib/Target/AMDGPU/AMDGPUFeatures.td
index 6ca896c..db00f8f 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUFeatures.td
+++ b/llvm/lib/Target/AMDGPU/AMDGPUFeatures.td
@@ -29,16 +29,16 @@ def FeatureLocalMemorySize0 : SubtargetFeatureLocalMemorySize<0>;
 def FeatureLocalMemorySize32768 : SubtargetFeatureLocalMemorySize<32768>;
 def FeatureLocalMemorySize65536 : SubtargetFeatureLocalMemorySize<65536>;
 
-class SubtargetFeatureWavefrontSize <int Value> : SubtargetFeature<
-  "wavefrontsize"#Value,
-  "WavefrontSize",
-  !cast<string>(Value),
+class SubtargetFeatureWavefrontSize <int ValueLog2> : SubtargetFeature<
+  "wavefrontsize"#!shl(1, ValueLog2),
+  "WavefrontSizeLog2",
+  !cast<string>(ValueLog2),
   "The number of threads per wavefront"
 >;
 
-def FeatureWavefrontSize16 : SubtargetFeatureWavefrontSize<16>;
-def FeatureWavefrontSize32 : SubtargetFeatureWavefrontSize<32>;
-def FeatureWavefrontSize64 : SubtargetFeatureWavefrontSize<64>;
+def FeatureWavefrontSize16 : SubtargetFeatureWavefrontSize<4>;
+def FeatureWavefrontSize32 : SubtargetFeatureWavefrontSize<5>;
+def FeatureWavefrontSize64 : SubtargetFeatureWavefrontSize<6>;
 
 class SubtargetFeatureGeneration <string Value, string FeatureName,
                                  string Subtarget,
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp b/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
index 3e24c4a..bd9cb92 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.cpp
@@ -126,8 +126,8 @@ GCNSubtarget::initializeSubtargetDependencies(const Triple &TT,
   }
 
   // Don't crash on invalid devices.
-  if (WavefrontSize == 0)
-    WavefrontSize = 64;
+  if (WavefrontSizeLog2 == 0)
+    WavefrontSizeLog2 = 5;
 
   HasFminFmaxLegacy = getGeneration() < AMDGPUSubtarget::VOLCANIC_ISLANDS;
 
@@ -163,7 +163,7 @@ AMDGPUSubtarget::AMDGPUSubtarget(const Triple &TT) :
   HasTrigReducedRange(false),
   MaxWavesPerEU(10),
   LocalMemorySize(0),
-  WavefrontSize(0)
+  WavefrontSizeLog2(0)
   { }
 
 GCNSubtarget::GCNSubtarget(const Triple &TT, StringRef GPU, StringRef FS,
diff --git a/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.h b/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.h
index 4db2a0f..604a766 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.h
+++ b/llvm/lib/Target/AMDGPU/AMDGPUSubtarget.h
@@ -77,7 +77,7 @@ protected:
   bool HasTrigReducedRange;
   unsigned MaxWavesPerEU;
   int LocalMemorySize;
-  unsigned WavefrontSize;
+  char WavefrontSizeLog2;
 
 public:
   AMDGPUSubtarget(const Triple &TT);
@@ -181,7 +181,11 @@ public:
   }
 
   unsigned getWavefrontSize() const {
-    return WavefrontSize;
+    return 1 << WavefrontSizeLog2;
+  }
+
+  unsigned getWavefrontSizeLog2() const {
+    return WavefrontSizeLog2;
   }
 
   int getLocalMemorySize() const {
@@ -237,8 +241,8 @@ public:
   /// \returns Corresponsing DWARF register number mapping flavour for the
   /// \p WavefrontSize.
   AMDGPUDwarfFlavour getAMDGPUDwarfFlavour() const {
-    return WavefrontSize == 32 ? AMDGPUDwarfFlavour::Wave32
-                               : AMDGPUDwarfFlavour::Wave64;
+    return getWavefrontSize() == 32 ? AMDGPUDwarfFlavour::Wave32
+                                    : AMDGPUDwarfFlavour::Wave64;
   }
 
   virtual ~AMDGPUSubtarget() {}
@@ -454,10 +458,6 @@ public:
     return (Generation)Gen;
   }
 
-  unsigned getWavefrontSizeLog2() const {
-    return Log2_32(WavefrontSize);
-  }
-
   /// Return the number of high bits known to be zero fror a frame index.
   unsigned getKnownHighZeroBitsForFrameIndex() const {
     return countLeadingZeros(MaxWaveScratchSize) + getWavefrontSizeLog2();
@@ -1160,7 +1160,7 @@ public:
       const override;
 
   bool isWave32() const {
-    return WavefrontSize == 32;
+    return getWavefrontSize() == 32;
   }
 
   const TargetRegisterClass *getBoolRC() const {