[llvm-commits] CVS: llvm/lib/Transforms/Scalar/LoopUnswitch.cpp

[Owen Anderson]

Changes in directory llvm/lib/Transforms/Scalar:

LoopUnswitch.cpp updated: 1.43 -> 1.44
---
Log message:

Switch to a very conservative heuristic for determining when loop-unswitching
will be profitable.  This is mainly to remove some cases where excessive
unswitching would result in long compile times and/or huge generated code.

Once someone comes up with a better heuristic that avoids these cases, this
should be switched out.


---
Diffs of the changes:  (+5 -5)

 LoopUnswitch.cpp |   10 +++++-----
 1 files changed, 5 insertions(+), 5 deletions(-)


Index: llvm/lib/Transforms/Scalar/LoopUnswitch.cpp
diff -u llvm/lib/Transforms/Scalar/LoopUnswitch.cpp:1.43 llvm/lib/Transforms/Scalar/LoopUnswitch.cpp:1.44
--- llvm/lib/Transforms/Scalar/LoopUnswitch.cpp:1.43	Wed Jun 28 11:38:55 2006
+++ llvm/lib/Transforms/Scalar/LoopUnswitch.cpp	Wed Jun 28 12:47:50 2006
@@ -333,11 +333,11 @@
   if (IsTrivialUnswitchCondition(L, LIC))
     return 0;
   
-  // If the loop is really large (over twice our threshold) don't even consider
-  // unswitching it.  This will produce a really large loop with lots of empty
-  // blocks.
-  if (L->getBlocks().size() > 2*Threshold)
-    return 2*Threshold;
+  // FIXME: This is really overly conservative.  However, more liberal 
+  // estimations have thus far resulted in excessive unswitching, which is bad
+  // both in compile time and in code size.  This should be replaced once
+  // someone figures out how a good estimation.
+  return L->getBlocks().size();
   
   unsigned Cost = 0;
   // FIXME: this is brain dead.  It should take into consideration code