2005-10-18 |
Chris Lattner | Add an option to this pass. If it is set, we are allow... |
tree | commitdiff |
2005-10-18 |
Chris Lattner | Fold (select C, load A, load B) -> load (select C,... |
tree | commitdiff |
2005-10-18 |
Nate Begeman | Do the right thing and enable 64 bit regs under the... |
tree | commitdiff |
2005-10-18 |
Nate Begeman | First bits of 64 bit PowerPC stuff, currently disabled... |
tree | commitdiff |
2005-10-18 |
Nate Begeman | Implement some feedback from Chris re: constant canonic... |
tree | commitdiff |
2005-10-18 |
Nate Begeman | Legalize BUILD_PAIR appropriately for upcoming 64 bit... |
tree | commitdiff |
2005-10-17 |
Nate Begeman | fold fmul X, +2.0 -> fadd X, X; |
tree | commitdiff |
2005-10-17 |
Chris Lattner | Make this work for FP constantexprs |
tree | commitdiff |
2005-10-17 |
Chris Lattner | Oops, X+0.0 isn't foldable, but X+-0.0 is. |
tree | commitdiff |
2005-10-17 |
Chris Lattner | relax this a bit, as we only support the default roundi... |
tree | commitdiff |
2005-10-17 |
Chris Lattner | add a trivial fold |
tree | commitdiff |
2005-10-16 |
Nate Begeman | More PPC32 -> PPC changes, as well as merging some... |
tree | commitdiff |
2005-10-15 |
Chris Lattner | Fix this logic. |
tree | commitdiff |
2005-10-15 |
Chris Lattner | Add a case we were missing that was causing us to fail... |
tree | commitdiff |
2005-10-15 |
Nate Begeman | Remove some dead code now that the dag combiner exists. |
tree | commitdiff |
2005-10-15 |
Chris Lattner | Remove some dead code: the ORI/ORIS cases are autogen... |
tree | commitdiff |
2005-10-15 |
Chris Lattner | prune #includes |
tree | commitdiff |
2005-10-15 |
Chris Lattner | These instructions are now autogenerated |
tree | commitdiff |
2005-10-15 |
Chris Lattner | Add a pattern for FSQRTS |
tree | commitdiff |
2005-10-15 |
Chris Lattner | remove dead code |
tree | commitdiff |
2005-10-15 |
Chris Lattner | Use getExtLoad here instead of getNode, as extloads... |
tree | commitdiff |
2005-10-15 |
Chris Lattner | remove broken SRA/rlwimi case |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Rename PPC32*.h to PPC*.h |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Merge PPCJITInfo.h and PPC32JITInfo.h. Note that the... |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Rename PowerPC*.h to PPC*.h |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Rename PowerPCInstrBuilder.h -> PPC* |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Nuke the PowerPCTargetMachine.h header. Note that... |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Rename PowerPC*.td -> PPC*.td |
tree | commitdiff |
2005-10-14 |
Chris Lattner | These are dead |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Eliminate PowerPC.td and PPC32.td, consolidating them... |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Like the comment says... |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Nuke PowerPCInstrFormats.h, its contents are dead.... |
tree | commitdiff |
2005-10-14 |
Nate Begeman | Properly split f32 and f64 into separate register class... |
tree | commitdiff |
2005-10-14 |
Nate Begeman | Remove an unnecsesary file. PPC32 and PPC64 share... |
tree | commitdiff |
2005-10-14 |
Chris Lattner | add the integer truncate/extension operations |
tree | commitdiff |
2005-10-14 |
Chris Lattner | These are now autogenerated |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Add patterns for FP round/extend |
tree | commitdiff |
2005-10-14 |
Chris Lattner | add a new SDTCisOpSmallerThanOp type constraint, and... |
tree | commitdiff |
2005-10-14 |
Nate Begeman | fold sext_in_reg, sext_in_reg where both have the same... |
tree | commitdiff |
2005-10-14 |
Chris Lattner | Allow $ |
tree | commitdiff |
2005-10-14 |
Nate Begeman | Relax the checking on zextload generation a bit, since... |
tree | commitdiff |
2005-10-13 |
Chris Lattner | Fix the trunc(load) case, finally allowing crafty and... |
tree | commitdiff |
2005-10-13 |
Chris Lattner | Fix some bugs in (sext (load x)) |
tree | commitdiff |
2005-10-13 |
Chris Lattner | When ExpandOp'ing a [SZ]EXTLOAD, make sure to remember... |
tree | commitdiff |
2005-10-13 |
Chris Lattner | Implement PromoteOp for *EXTLOAD, allowing MallocBench... |
tree | commitdiff |
2005-10-13 |
Nate Begeman | Fix the remaining DAGCombiner issues pointed out by... |
tree | commitdiff |
2005-10-13 |
Chris Lattner | Fix a minor bug in the dag combiner that broke pcompres... |
tree | commitdiff |
2005-10-13 |
Nate Begeman | Add support to Legalize for expanding i64 sextload... |
tree | commitdiff |
2005-10-13 |
Jim Laskey | Inhibit instructions from being pushed before function... |
tree | commitdiff |
2005-10-13 |
Nate Begeman | Move some Legalize functionality over to the DAGCombine... |
tree | commitdiff |
2005-10-12 |
Nate Begeman | Fix a potential bug with two combine-to's back to back... |
tree | commitdiff |
2005-10-12 |
Nate Begeman | More cool stuff for the dag combiner. We can now final... |
tree | commitdiff |
2005-10-12 |
Jim Laskey | Finally committing to the new scheduler. Still -sched... |
tree | commitdiff |
2005-10-12 |
Jim Laskey | Added graphviz/gv support for MF. |
tree | commitdiff |
2005-10-11 |
Chris Lattner | Fix (hopefully the last) issue where LSR is nondetermin... |
tree | commitdiff |
2005-10-11 |
Chris Lattner | Fix another problem where LSR was being nondeterminstic... |
tree | commitdiff |
2005-10-11 |
Chris Lattner | Fix another lsr-is-nondeterministic case |
tree | commitdiff |
2005-10-11 |
Chris Lattner | Fix a powerpc crash on CodeGen/Generic/llvm-ct-intrinsi... |
tree | commitdiff |
2005-10-11 |
Chris Lattner | Add a canonicalization that got lost, fixing PowerPC... |
tree | commitdiff |
2005-10-10 |
Chris Lattner | clean up some corner cases |
tree | commitdiff |
2005-10-10 |
Chris Lattner | Implement trivial DSE. If two stores are neighbors... |
tree | commitdiff |
2005-10-10 |
Chris Lattner | Add support for CombineTo, allowing the dag combiner... |
tree | commitdiff |
2005-10-10 |
Nate Begeman | Teach the DAGCombiner several new tricks, teaching... |
tree | commitdiff |
2005-10-10 |
Chris Lattner | Fix comment |
tree | commitdiff |
2005-10-10 |
Chris Lattner | Add ISD::ADD to MaskedValueIsZero |
tree | commitdiff |
2005-10-10 |
Chris Lattner | This function is now dead |
tree | commitdiff |
2005-10-10 |
Chris Lattner | Enable Nate's excellent DAG combiner work by default... |
tree | commitdiff |
2005-10-10 |
Chris Lattner | These definitions have been moved to common code. |
tree | commitdiff |
2005-10-10 |
Chris Lattner | Pull DAG ISel generation nodes out of the PowerPC backe... |
tree | commitdiff |
2005-10-09 |
Chris Lattner | add a todo for something I noticed |
tree | commitdiff |
2005-10-09 |
Chris Lattner | (X & Y) & C == 0 if either X&C or Y&C are zero |
tree | commitdiff |
2005-10-09 |
Chris Lattner | Make MaskedValueIsZero a bit more aggressive |
tree | commitdiff |
2005-10-09 |
Andrew Lenharth | This seems useful from the original patch that added... |
tree | commitdiff |
2005-10-09 |
Chris Lattner | Fix funky xcode indentation |
tree | commitdiff |
2005-10-09 |
Chris Lattner | Hrm, you didn't see this. |
tree | commitdiff |
2005-10-09 |
Chris Lattner | Fix a source of non-determinism in the backend: the... |
tree | commitdiff |
2005-10-09 |
Chris Lattner | When emiting a CopyFromReg and the source is already... |
tree | commitdiff |
2005-10-09 |
Chris Lattner | Disable formation of rlwinm instructions from SRA bases... |
tree | commitdiff |
2005-10-08 |
Nate Begeman | Remove another unused file. Preparing for the great... |
tree | commitdiff |
2005-10-08 |
Nate Begeman | Remove a file that is no longer used |
tree | commitdiff |
2005-10-08 |
Nate Begeman | Lo and behold, the last bits of SelectionDAG.cpp have... |
tree | commitdiff |
2005-10-07 |
Chris Lattner | When preselecting, favor things that have low depth... |
tree | commitdiff |
2005-10-07 |
Chris Lattner | remove debugging code |
tree | commitdiff |
2005-10-07 |
Chris Lattner | implement CodeGen/PowerPC/div-2.ll:test2-4 by propagati... |
tree | commitdiff |
2005-10-07 |
Chris Lattner | fix indentation |
tree | commitdiff |
2005-10-07 |
Chris Lattner | Turn sdivs into udivs when we can prove the sign bits... |
tree | commitdiff |
2005-10-07 |
Jeff Cohen | Remove useless variable. |
tree | commitdiff |
2005-10-07 |
Chris Lattner | add a hack to work around broken VC++ scoping rules... |
tree | commitdiff |
2005-10-07 |
Chris Lattner | Fix a CQ regression from my patch to split F32/F64... |
tree | commitdiff |
2005-10-07 |
Chris Lattner | Fix CodeGen/Generic/bool-to-double.ll |
tree | commitdiff |
2005-10-06 |
Chris Lattner | Pull out Call, reducing stack frame size from 6032... |
tree | commitdiff |
2005-10-06 |
Chris Lattner | Pull out setcc, this reduces stack frame size from... |
tree | commitdiff |
2005-10-06 |
Chris Lattner | Pull two more methods out, reducing stack frame size... |
tree | commitdiff |
2005-10-06 |
Chris Lattner | Add a recursive-iterative hybrid stage to attempt to... |
tree | commitdiff |
2005-10-06 |
Chris Lattner | silence a bogus GCC warning |
tree | commitdiff |
2005-10-06 |
Chris Lattner | Fix the LLC regressions on X86 last night. In particul... |
tree | commitdiff |
2005-10-06 |
Andrew Lenharth | This is suppose to work now |
tree | commitdiff |
2005-10-06 |
Andrew Lenharth | remove VAX compatibility instruction, we will never... |
tree | commitdiff |
2005-10-06 |
Chris Lattner | Make the legalizer completely non-recursive |
tree | commitdiff |
2005-10-05 |
Nate Begeman | Let the combiner handle more cases |
tree | commitdiff |
next |