| Commit message (Collapse) | Author | Age | Files | Lines | |
|---|---|---|---|---|---|
| * | ADd two new 'add' methods | Chris Lattner | 2003-03-03 | 1 | -0/+28 |
| | | | | | llvm-svn: 5691 | ||||
| * | Don't apply type information to load instructions if it will cause collapsing | Chris Lattner | 2003-03-03 | 2 | -6/+7 |
| | | | | | llvm-svn: 5684 | ||||
| * | Fix a problem with negative indexes | Chris Lattner | 2003-03-02 | 1 | -3/+3 |
| | | | | | llvm-svn: 5681 | ||||
| * | Add dump method for Loops | Chris Lattner | 2003-02-28 | 1 | -0/+4 |
| | | | | | llvm-svn: 5671 | ||||
| * | Simplify a bit by using a new member function | Chris Lattner | 2003-02-27 | 1 | -6/+3 |
| | | | | | llvm-svn: 5662 | ||||
| * | Change behavior of changeExitBlock function to replace all instances of exit ↵ | Chris Lattner | 2003-02-27 | 1 | -2/+8 |
| | | | | | | | block llvm-svn: 5661 | ||||
| * | Fix bug: LICM/2003-02-27-PreheaderProblem.ll | Chris Lattner | 2003-02-27 | 1 | -0/+7 |
| | | | | | | | | | There may be a single outside predecessor and still need a new loop-preheader if the predecessor has multiple successors. llvm-svn: 5656 | ||||
| * | - LoopInfo now calculates and tracks loop exit blocks | Chris Lattner | 2003-02-27 | 1 | -11/+40 |
| | | | | | llvm-svn: 5650 | ||||
| * | Make the aliassettracker much more precise by actually tracking size | Chris Lattner | 2003-02-26 | 1 | -20/+33 |
| | | | | | | | information for various accesses. What a concept. llvm-svn: 5647 | ||||
| * | Fix bug: BasicAA/2003-02-26-AccessSizeTest.ll | Chris Lattner | 2003-02-26 | 1 | -11/+36 |
| | | | | | llvm-svn: 5645 | ||||
| * | Fix several bugs in basic-aa | Chris Lattner | 2003-02-26 | 1 | -5/+10 |
| | | | | | llvm-svn: 5643 | ||||
| * | Add new -no-aa implementation | Chris Lattner | 2003-02-26 | 1 | -0/+19 |
| | | | | | llvm-svn: 5641 | ||||
| * | Move BasicAA pass out to it's own header file | Chris Lattner | 2003-02-26 | 2 | -258/+292 |
| | | | | | llvm-svn: 5640 | ||||
| * | Adjust to implement new AA interface | Chris Lattner | 2003-02-26 | 2 | -31/+17 |
| | | | | | llvm-svn: 5638 | ||||
| * | Adjust to new AA interface | Chris Lattner | 2003-02-26 | 1 | -1/+1 |
| | | | | | llvm-svn: 5637 | ||||
| * | Convert to work with new AliasAnalysis interface by conservatively assuming ↵ | Chris Lattner | 2003-02-26 | 1 | -6/+8 |
| | | | | | | | all pointers are arbitrarily large accesses llvm-svn: 5636 | ||||
| * | Calculate and pass load sizes to the alias analysis infrastructure | Chris Lattner | 2003-02-26 | 1 | -8/+19 |
| | | | | | llvm-svn: 5635 | ||||
| * | Adjust to new AA interface | Chris Lattner | 2003-02-26 | 1 | -17/+48 |
| | | | | | | | Add tracking for Mod/Ref info llvm-svn: 5634 | ||||
| * | - Checkin of the alias analysis work: | Chris Lattner | 2003-02-26 | 1 | -78/+187 |
| | | | | | | | | | | * Takes into account the size of the memory reference to determine aliasing. * Expose mod/ref information in a more consistent way * BasicAA can now disambiguate A[i][1] and A[j][2] for conservative request sizes llvm-svn: 5633 | ||||
| * | This is a substantial rewrite of the AliasSetTracker class which now uses | Chris Lattner | 2003-02-24 | 1 | -111/+218 |
| | | | | | | | | | | | | a union-find based algorithm, is significantly faster, and is more general. It will also scale to handle call instructions correctly, which is a nice added bonus. This includes a new pass -print-alias-sets which can be used to show how alias sets are formed for a particular analysis. llvm-svn: 5619 | ||||
| * | Dramatically simplify building of natural loops and fix a bug where the BBMap | Chris Lattner | 2003-02-22 | 1 | -37/+23 |
| | | | | | | | was not correctly computed. llvm-svn: 5606 | ||||
| * | Fix the requisite bug that I introduced | Chris Lattner | 2003-02-20 | 1 | -11/+13 |
| | | | | | llvm-svn: 5605 | ||||
| * | Fix 80 character formatting | Chris Lattner | 2003-02-20 | 1 | -4/+4 |
| | | | | | llvm-svn: 5604 | ||||
| * | Fix bug: 2003-02-19-LoopInfoNestingBug.ll | Chris Lattner | 2003-02-20 | 1 | -15/+35 |
| | | | | | llvm-svn: 5603 | ||||
| * | Fix resolution of indirect function calls... whoops | Chris Lattner | 2003-02-14 | 1 | -1/+1 |
| | | | | | llvm-svn: 5576 | ||||
| * | Mark function edges | Chris Lattner | 2003-02-14 | 1 | -0/+2 |
| | | | | | llvm-svn: 5571 | ||||
| * | - Eliminate provably non-pointer nodes from graphs. | Chris Lattner | 2003-02-14 | 1 | -5/+22 |
| | | | | | | | | | | | | | | | This helps a lot of testcases, for example: New Time New #Nodes Old Time Old #Nodes 254.gap: 91.1024 21605 91.1397 22657 povray31: 2.7807 8613 3.0152 10338 255.vortex: 1.2034 8153 1.2172 8822 moria: .6756 3150 .7054 3877 300.twolf: .1652 2010 .1851 3270 Typically, testcases which use long and ulong integers a lot get better, f.e. povray above. llvm-svn: 5566 | ||||
| * | Don't put integer pointers (longs) into the scalar map. | Chris Lattner | 2003-02-14 | 1 | -0/+10 |
| | | | | | | | This speeds stuff up by 10% on some tests, woot! llvm-svn: 5564 | ||||
| * | Hax0r around a deficiency in the Pass infrastructure | Chris Lattner | 2003-02-13 | 1 | -2/+5 |
| | | | | | llvm-svn: 5555 | ||||
| * | Use the new tailclip property new added to graphviz CVS to make call nodes | Chris Lattner | 2003-02-13 | 1 | -5/+4 |
| | | | | | | | actually intelligable llvm-svn: 5554 | ||||
| * | Move node forwarding code from being inlined to being out-of-line. | Chris Lattner | 2003-02-13 | 1 | -0/+21 |
| | | | | | | | This brings a 11.6% speedup to steens, and a 3.6 overall speedup to ds-aa llvm-svn: 5552 | ||||
| * | Rename options | Chris Lattner | 2003-02-11 | 1 | -3/+3 |
| | | | | | llvm-svn: 5537 | ||||
| * | Implement a "union-findy" version of DS-Analysis, which eliminates the | Chris Lattner | 2003-02-11 | 4 | -238/+126 |
| | | | | | | | Referrers list on DSNodes. llvm-svn: 5536 | ||||
| * | Minor tweaks to printing. Close the file before viewing it in viewGraph | Chris Lattner | 2003-02-11 | 1 | -1/+3 |
| | | | | | llvm-svn: 5535 | ||||
| * | Modest speedup which seems to help steens quite a bit on large graphs | Chris Lattner | 2003-02-10 | 1 | -1/+2 |
| | | | | | llvm-svn: 5532 | ||||
| * | Fix problem breaking GAP, use hasNoReferrers more | Chris Lattner | 2003-02-10 | 1 | -4/+4 |
| | | | | | llvm-svn: 5530 | ||||
| * | Implement a new "viewGraph" method which can be used to instantly view a ↵ | Chris Lattner | 2003-02-10 | 1 | -0/+17 |
| | | | | | | | graph from gdb llvm-svn: 5528 | ||||
| * | Rearrange code | Chris Lattner | 2003-02-10 | 1 | -11/+13 |
| | | | | | llvm-svn: 5527 | ||||
| * | Fix problem breaking Fhourstones | Chris Lattner | 2003-02-10 | 1 | -1/+2 |
| | | | | | llvm-svn: 5526 | ||||
| * | Make steensgaards performance not shameful | Chris Lattner | 2003-02-10 | 1 | -6/+12 |
| | | | | | llvm-svn: 5524 | ||||
| * | Don't insert null entries into the scalar map for constexpr (cast null to Ty) | Chris Lattner | 2003-02-09 | 1 | -5/+11 |
| | | | | | llvm-svn: 5523 | ||||
| * | Allow printing of various types of alias analysis results | Chris Lattner | 2003-02-09 | 1 | -4/+27 |
| | | | | | llvm-svn: 5520 | ||||
| * | Implement knowledge in BasicAA that &A->field != &A and (P+1) != P | Chris Lattner | 2003-02-09 | 1 | -0/+17 |
| | | | | | llvm-svn: 5519 | ||||
| * | - Fix BasicAA to correctly detect the non-aliasness of A[1] & A[2] | Chris Lattner | 2003-02-09 | 1 | -2/+2 |
| | | | | | llvm-svn: 5518 | ||||
| * | Switch to using the right call vector | Chris Lattner | 2003-02-09 | 1 | -0/+2 |
| | | | | | llvm-svn: 5517 | ||||
| * | Fix major bugs in incompleteness marking that were pessimizing results | Chris Lattner | 2003-02-09 | 1 | -96/+101 |
| | | | | | llvm-svn: 5515 | ||||
| * | Print out the steens graph if -debug is specified | Chris Lattner | 2003-02-09 | 1 | -1/+2 |
| | | | | | llvm-svn: 5514 | ||||
| * | * Fix a bug where global incompleteness marking would not mark the global | Chris Lattner | 2003-02-09 | 1 | -8/+5 |
| | | | | | | | | | itself incomplete! * Allow incompleteness callers to specify they don't want globals to be considered sources of incompleteness. llvm-svn: 5513 | ||||
| * | Add a special case for main because we know it's incoming arguments don't alias | Chris Lattner | 2003-02-08 | 1 | -1/+1 |
| | | | | | llvm-svn: 5511 | ||||
| * | Fix a bug where we would mark the callees arguments incomplete if the caller ↵ | Chris Lattner | 2003-02-08 | 1 | -2/+2 |
| | | | | | | | was external. llvm-svn: 5510 | ||||

