Switch-to-lookup tables: set threshold to 3 cases

There has been an old FIXME to find the right cut-off for when it's worth
analyzing and potentially transforming a switch to a lookup table.

The switches always have two or more cases. I could not measure any speed-up
by transforming a switch with two cases. A switch with three cases gets a nice
speed-up, and I couldn't measure any compile-time regression, so I think this
is the right threshold.

In a Clang self-host, this causes 480 new switches to be transformed,
and reduces the final binary size with 8 KB.

llvm-svn: 199294

1 files changed, 3 insertions, 5 deletions

diff --git a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
index 90f8847bc1c..a30dcf2fe0c 100644
--- a/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
+++ b/llvm/lib/Transforms/Utils/SimplifyCFG.cpp
@@ -3684,11 +3684,9 @@ static bool SwitchToLookupTable(SwitchInst *SI,
   // GEP needs a runtime relocation in PIC code. We should just build one big
   // string and lookup indices into that.
 
-  // Ignore the switch if the number of cases is too small.
-  // This is similar to the check when building jump tables in
-  // SelectionDAGBuilder::handleJTSwitchCase.
-  // FIXME: Determine the best cut-off.
-  if (SI->getNumCases() < 4)
+  // Ignore switches with less than three cases. Lookup tables will not make them
+  // faster, so we don't analyze them.
+  if (SI->getNumCases() < 3)
     return false;
 
   // Figure out the corresponding result for each case value and phi node in the