Remove floating point computations form SpillPlacement.cpp.

Patch by Benjamin Kramer!

Use the BlockFrequency class instead of floats in the Hopfield network
computations. This rescales the node Bias field from a [-2;2] float
range to two block frequencies BiasN and BiasP pulling in opposite
directions. This construct has a more predictable behavior when block
frequencies saturate.

The per-node scaling factors are no longer necessary, assuming the block
frequencies around a bundle are consistent.

This patch can cause the register allocator to make different spilling
decisions. The differences should be small.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@185393 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 3 insertions, 0 deletions

diff --git a/include/llvm/Support/BlockFrequency.h b/include/llvm/Support/BlockFrequency.h
index 147d52a..624d6de 100644
--- a/include/llvm/Support/BlockFrequency.h
+++ b/include/llvm/Support/BlockFrequency.h
@@ -36,6 +36,9 @@ public:
   /// \brief Returns the frequency of the entry block of the function.
   static uint64_t getEntryFrequency() { return ENTRY_FREQ; }
 
+  /// \brief Returns the maximum possible prequency, the saturation value.
+  static uint64_t getMaxFrequency() { return -1ULL; }
+
   /// \brief Returns the frequency as a fixpoint number scaled by the entry
   /// frequency.
   uint64_t getFrequency() const { return Frequency; }