llvm-6502

mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-10-04 00:54:53 +00:00

History

Dan Gohman 948c8a3e3d When checking whether the special handling for an addrec increment which doesn't dominate the header is needed, don't check whether the increment expression has computable loop evolution. While the operands of an addrec are required to be loop-invariant, they're not required to dominate any part of the loop. This fixes PR6914. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@102389 91177308-0d34-0410-b5e6-96231b3b80d8		2010-04-26 21:46:36 +00:00
..
IPA	fix callgraph dump to not print 0x0x1234 for nodes.	2010-04-23 18:23:40 +00:00
AliasAnalysis.cpp
AliasAnalysisCounter.cpp
AliasAnalysisEvaluator.cpp	clean up algorithm and remove operand order assumptions	2010-04-08 16:46:24 +00:00
AliasDebugger.cpp
AliasSetTracker.cpp
Analysis.cpp
BasicAliasAnalysis.cpp	Revert 101465, it broke internal OpenGL testing.	2010-04-16 23:37:20 +00:00
CaptureTracking.cpp
CFGPrinter.cpp
CMakeLists.txt	Update CMake build.	2010-04-08 18:52:18 +00:00
ConstantFolding.cpp	Revert 101465, it broke internal OpenGL testing.	2010-04-16 23:37:20 +00:00
DbgInfoPrinter.cpp
DebugInfo.cpp	DebugInfoFinder::processModule was foiling my plot by	2010-04-02 20:44:29 +00:00
DomPrinter.cpp	Remove dead code in the dotty dominance tree printer.	2010-04-12 15:02:19 +00:00
InlineCost.cpp	Revert r101471. For tight recursive functions which have multiple	2010-04-21 00:43:30 +00:00
InstCount.cpp
InstructionSimplify.cpp	move some select simplifications out out instcombine into	2010-04-20 05:32:14 +00:00
Interval.cpp
IntervalPartition.cpp
IVUsers.cpp	Remove the Expr member from IVUsers. Instead of remembering the expression,	2010-04-19 21:48:58 +00:00
LazyValueInfo.cpp
LibCallAliasAnalysis.cpp
LibCallSemantics.cpp
Lint.cpp	Tidy a comment.	2010-04-22 01:30:05 +00:00
LiveValues.cpp
LoopDependenceAnalysis.cpp
LoopInfo.cpp	Fix a bunch of namespace polution.	2010-04-15 17:08:50 +00:00
LoopPass.cpp	Ok, third time's the charm. No changes from last time except the CMake	2010-04-02 23:17:14 +00:00
Makefile
MemoryBuiltins.cpp	Revert 101465, it broke internal OpenGL testing.	2010-04-16 23:37:20 +00:00
MemoryDependenceAnalysis.cpp	Revert 101465, it broke internal OpenGL testing.	2010-04-16 23:37:20 +00:00
PHITransAddr.cpp
PointerTracking.cpp	Rename isLoopGuardedByCond to isLoopEntryGuardedByCond, to emphasise	2010-04-11 19:27:13 +00:00
PostDominators.cpp	Remove unneeded debug in PostDominator runOnFunction()	2010-04-12 15:32:55 +00:00
ProfileEstimatorPass.cpp
ProfileInfo.cpp
ProfileInfoLoader.cpp
ProfileInfoLoaderPass.cpp
ProfileVerifierPass.cpp
README.txt	When checking whether the special handling for an addrec increment which	2010-04-26 21:46:36 +00:00
ScalarEvolution.cpp	ScalarEvolution support for <= and >= loops.	2010-04-24 03:09:42 +00:00
ScalarEvolutionAliasAnalysis.cpp
ScalarEvolutionExpander.cpp	When checking whether the special handling for an addrec increment which	2010-04-26 21:46:36 +00:00
ScalarEvolutionNormalization.cpp	Generalize IVUsers to track arbitrary expressions rather than expressions	2010-04-07 22:27:08 +00:00
SparsePropagation.cpp
Trace.cpp
ValueTracking.cpp	Revert 101465, it broke internal OpenGL testing.	2010-04-16 23:37:20 +00:00

README.txt

Analysis Opportunities:

//===---------------------------------------------------------------------===//

In test/Transforms/LoopStrengthReduce/quadradic-exit-value.ll, the
ScalarEvolution expression for %r is this:

  {1,+,3,+,2}<loop>

Outside the loop, this could be evaluated simply as (%n * %n), however
ScalarEvolution currently evaluates it as

  (-2 + (2 * (trunc i65 (((zext i64 (-2 + %n) to i65) * (zext i64 (-1 + %n) to i65)) /u 2) to i64)) + (3 * %n))

In addition to being much more complicated, it involves i65 arithmetic,
which is very inefficient when expanded into code.

//===---------------------------------------------------------------------===//

In formatValue in test/CodeGen/X86/lsr-delayed-fold.ll,

ScalarEvolution is forming this expression:

((trunc i64 (-1 * %arg5) to i32) + (trunc i64 %arg5 to i32) + (-1 * (trunc i64 undef to i32)))

This could be folded to

(-1 * (trunc i64 undef to i32))

//===---------------------------------------------------------------------===//