CodeGen: BlockPlacement: Reduce TriangleChainCount to 2

[iteratee]

Summary

This produces a 1% speedup on an important internal Google benchmark
(protocol buffers), with no other regressions in google or in the llvm
test-suite. Only 5 targets in the entire llvm test-suite are affected,
and on those 5 targets the size increase is 0.027%

[davidxl]

lgtm

[hfinkel]

Could you make a test case? Could be an MIR test case if an IR case would be too fragile.

[echristo]

In D31013#702483, @hfinkel wrote:

Could you make a test case? Could be an MIR test case if an IR case would be too fragile.

Not necessarily disagreeing, but MIR and "not too fragile"? :)

[hfinkel]

In D31013#702507, @echristo wrote:

In D31013#702483, @hfinkel wrote:

Could you make a test case? Could be an MIR test case if an IR case would be too fragile.

Not necessarily disagreeing, but MIR and "not too fragile"? :)

Yea, okay :) -- In any case, it looks like this was committed in r297925 and there were indeed some test-case changes.

[iteratee]

Committed in rL297925

lib/CodeGen/MachineBlockPlacement.cpp

static cl::opt<unsigned> TailDupPlacementPenalty(
     cl::init(2),
     cl::Hidden);
 
 // Heuristic for triangle chains.
 static cl::opt<unsigned> TriangleChainCount(
     "triangle-chain-count",
     cl::desc("Number of triangle-shaped-CFG's that need to be in a row for the "
              "triangle tail duplication heuristic to kick in. 0 to disable."),
-    cl::init(3),
+    cl::init(2),
     cl::Hidden);
 
 extern cl::opt<unsigned> StaticLikelyProb;
 extern cl::opt<unsigned> ProfileLikelyProb;
 
 // Internal option used to control BFI display only after MBP pass.
 // Defined in CodeGen/MachineBlockFrequencyInfo.cpp:
 // -view-block-layout-with-bfi=