llvm-6502

mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-12-27 13:30:05 +00:00

History

Paul Robinson cf84b5ba5e Constify the Optnone checks in IR passes. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@202213 91177308-0d34-0410-b5e6-96231b3b80d8		2014-02-26 01:23:26 +00:00
..
IPA	Use DataLayout from the module when easily available.	2014-02-25 23:25:17 +00:00
AliasAnalysis.cpp	Make DataLayout a plain object, not a pass.	2014-02-25 17:30:31 +00:00
AliasAnalysisCounter.cpp
AliasAnalysisEvaluator.cpp
AliasDebugger.cpp
AliasSetTracker.cpp
Analysis.cpp	[PM] Make the verifier work independently of any pass manager.	2014-01-19 02:22:18 +00:00
BasicAliasAnalysis.cpp	Rename a few more DataLayout variables from TD to DL.	2014-02-21 18:34:28 +00:00
BlockFrequencyInfo.cpp
BranchProbabilityInfo.cpp
CaptureTracking.cpp	Make nocapture analysis work with addrspacecast	2014-01-14 19:11:52 +00:00
CFG.cpp	Make succ_iterator a real random access iterator and clean up a couple of users.	2014-02-10 14:17:42 +00:00
CFGPrinter.cpp	Replace the F_Binary flag with a F_Text one.	2014-02-24 18:20:12 +00:00
CMakeLists.txt	[PM] Add a new "lazy" call graph analysis pass for the new pass manager.	2014-02-06 04:37:03 +00:00
CodeMetrics.cpp
ConstantFolding.cpp
CostModel.cpp	Reduce code duplication resulting from the ConstantVector/ConstantDataVector split.	2014-02-13 16:48:38 +00:00
Delinearization.cpp
DependenceAnalysis.cpp	normalize the last delinearized dimension	2014-02-21 18:15:11 +00:00
DominanceFrontier.cpp
DomPrinter.cpp
InstCount.cpp
InstructionSimplify.cpp	Rename many DataLayout variables from TD to DL.	2014-02-21 00:06:31 +00:00
Interval.cpp
IntervalPartition.cpp
IVUsers.cpp	Make DataLayout a plain object, not a pass.	2014-02-25 17:30:31 +00:00
LazyCallGraph.cpp	[PM] Fix horrible typos that somehow didn't cause a failure in a C++11	2014-02-06 05:17:02 +00:00
LazyValueInfo.cpp	Make DataLayout a plain object, not a pass.	2014-02-25 17:30:31 +00:00
LibCallAliasAnalysis.cpp
LibCallSemantics.cpp
Lint.cpp	Make DataLayout a plain object, not a pass.	2014-02-25 17:30:31 +00:00
LLVMBuild.txt
Loads.cpp
LoopInfo.cpp
LoopPass.cpp	Constify the Optnone checks in IR passes.	2014-02-26 01:23:26 +00:00
Makefile
MemDepPrinter.cpp
MemoryBuiltins.cpp	Update optimization passes to handle inalloca arguments	2014-01-28 02:38:36 +00:00
MemoryDependenceAnalysis.cpp	Make DataLayout a plain object, not a pass.	2014-02-25 17:30:31 +00:00
ModuleDebugInfoPrinter.cpp
NoAliasAnalysis.cpp	Make DataLayout a plain object, not a pass.	2014-02-25 17:30:31 +00:00
PHITransAddr.cpp	Rename some member variables from TD to DL.	2014-02-18 15:33:12 +00:00
PostDominators.cpp
PtrUseVisitor.cpp	Hoist the GEP constant address offset computation to a common home on	2012-12-11 10:29:10 +00:00
README.txt
RegionInfo.cpp	[PM] Split DominatorTree into a concrete analysis result object which	2014-01-13 13:07:17 +00:00
RegionPass.cpp
RegionPrinter.cpp
ScalarEvolution.cpp	Make DataLayout a plain object, not a pass.	2014-02-25 17:30:31 +00:00
ScalarEvolutionAliasAnalysis.cpp
ScalarEvolutionExpander.cpp	Rename some member variables from TD to DL.	2014-02-18 15:33:12 +00:00
ScalarEvolutionNormalization.cpp
SparsePropagation.cpp
TargetTransformInfo.cpp	Make DataLayout a plain object, not a pass.	2014-02-25 17:30:31 +00:00
Trace.cpp	Put the functionality for printing a value to a raw_ostream as an	2014-01-09 02:29:41 +00:00
TypeBasedAliasAnalysis.cpp
ValueTracking.cpp	Allow speculating llvm.sqrt, fma and fmuladd	2014-01-31 00:09:00 +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))

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