Matthias Braun e152c1527d InstCombine: Restrict computeKnownBits() on all Values to OptLevel > 2 ... As part of r251146 InstCombine was extended to call computeKnownBits on every value in the function to determine whether it happens to be constant. This increases typical compiletime by 1-3% (5% in irgen+opt time) in my measurements. On the other hand this case did not trigger once in the whole llvm-testsuite. This patch introduces the notion of ExpensiveCombines which are only enabled for OptLevel > 2. I removed the check in InstructionSimplify as that is called from various places where the OptLevel is not known but given the rarity of the situation I think a check in InstCombine is enough. Differential Revision: http://reviews.llvm.org/D16835 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@263047 91177308-0d34-0410-b5e6-96231b3b80d8		2016-03-09 18:47:11 +00:00
..
assume.ll	[ValueTracking] do not overwrite analysis results already computed	2015-06-15 05:46:29 +00:00
known-bits-from-range-md.ll	InstCombine: Restrict computeKnownBits() on all Values to OptLevel > 2	2016-03-09 18:47:11 +00:00
known-non-equal.ll	[ValueTracking] Add a new predicate: isKnownNonEqual()	2015-10-22 13:18:42 +00:00
known-power-of-two.ll	[ValueTracking] fix bug computing isKnownToBeAPowerOfTwo() with arithmetic shift right (PR25900)	2015-12-30 22:40:52 +00:00
knownnonzero-shift.ll	[ValueTracking] Extend r251146 to catch a fairly common case	2015-10-26 14:10:46 +00:00
knownzero-shift.ll	[ValueTracking] Teach isKnownNonZero a new trick	2015-09-24 16:06:32 +00:00
memory-dereferenceable.ll	[gc.statepoint] Change gc.statepoint intrinsic's return type to token type instead of i32 type	2015-12-26 07:54:32 +00:00
monotonic-phi.ll	[ValueTracking] Teach isKnownNonZero about monotonically increasing PHIs	2015-09-29 14:08:45 +00:00
pr23011.ll