| 2010-10-29 |
Evan Cheng | Avoiding overly aggressive latency scheduling. If the...
|
commit | commitdiff | tree |
| 2010-10-29 |
Evan Cheng | - Don't schedule nodes with only MVT::Flag and MVT...
|
commit | commitdiff | tree |
| 2010-10-28 |
Evan Cheng | Re-commit 117518 and 117519 now that ARM MC test failures...
|
commit | commitdiff | tree |
| 2010-10-28 |
Evan Cheng | Disable most of the ARM vfp / NEON MC tests. These...
|
commit | commitdiff | tree |
| 2010-10-28 |
Evan Cheng | Revert 117518 and 117519 for now. They changed scheduling...
|
commit | commitdiff | tree |
| 2010-10-28 |
Evan Cheng | - Assign load / store with shifter op address modes...
|
commit | commitdiff | tree |
| 2010-10-28 |
Evan Cheng | Fix a major bug in operand latency computation. The...
|
commit | commitdiff | tree |
| 2010-10-27 |
Evan Cheng | Shifter ops are not always free. Do not fold them ...
|
commit | commitdiff | tree |
| 2010-10-27 |
Evan Cheng | Putting r117193 back except for the compile time cost...
|
commit | commitdiff | tree |
| 2010-10-26 |
Evan Cheng | Use instruction itinerary to determine what instructions...
|
commit | commitdiff | tree |
| 2010-10-26 |
Evan Cheng | NEON vmov's are in Neon domain.
|
commit | commitdiff | tree |
| 2010-10-25 |
Evan Cheng | Neuter r117193 as it causes significant post-ra scheduler...
|
commit | commitdiff | tree |
| 2010-10-23 |
Evan Cheng | Enable ARM fastcc.
|
commit | commitdiff | tree |
| 2010-10-23 |
Evan Cheng | Properly model the latency of register defs which are...
|
commit | commitdiff | tree |
| 2010-10-23 |
Evan Cheng | Latency between CPSR def and branch is zero.
|
commit | commitdiff | tree |
| 2010-10-22 |
Evan Cheng | Unbreak build.
|
commit | commitdiff | tree |
| 2010-10-22 |
Evan Cheng | Transfer implicit ops when forming load multiple and...
|
commit | commitdiff | tree |
| 2010-10-22 |
Evan Cheng | Silence compiler warnings.
|
commit | commitdiff | tree |
| 2010-10-22 |
Evan Cheng | Add fastcc cc: pass and return VFP / NEON values in...
|
commit | commitdiff | tree |
| 2010-10-21 |
Evan Cheng | Add missing scheduling itineraries for transfers between...
|
commit | commitdiff | tree |
| 2010-10-20 |
Evan Cheng | More accurate estimate / tracking of register pressure.
|
commit | commitdiff | tree |
| 2010-10-19 |
Evan Cheng | Re-enable register pressure aware machine licm with...
|
commit | commitdiff | tree |
| 2010-10-19 |
Evan Cheng | - Add a hook for target to determine whether an instruction...
|
commit | commitdiff | tree |
| 2010-10-16 |
Evan Cheng | More machine LICM work. It now tracks register pressure...
|
commit | commitdiff | tree |
| 2010-10-14 |
Evan Cheng | Register pressure and instruction latency aware machine...
|
commit | commitdiff | tree |
| 2010-10-13 |
Evan Cheng | Limit load / store issues (at least until we have a...
|
commit | commitdiff | tree |
| 2010-10-12 |
Evan Cheng | Turn some fp stackifier assertion into errors to avoid...
|
commit | commitdiff | tree |
| 2010-10-11 |
Evan Cheng | More ARM scheduling itinerary fixes.
|
commit | commitdiff | tree |
| 2010-10-11 |
Evan Cheng | Proper VST scheduling itineraries.
|
commit | commitdiff | tree |
| 2010-10-09 |
Evan Cheng | Add VLD4 scheduling itineraries.
|
commit | commitdiff | tree |
| 2010-10-09 |
Evan Cheng | Finish vld3 and vld4.
|
commit | commitdiff | tree |
| 2010-10-09 |
Evan Cheng | Complete vld2 instruction itineries.
|
commit | commitdiff | tree |
| 2010-10-09 |
Evan Cheng | Multiply instructions are issued on pipeline 0. They...
|
commit | commitdiff | tree |
| 2010-10-09 |
Evan Cheng | Correct some load / store instruction itinerary mistakes:
|
commit | commitdiff | tree |
| 2010-10-08 |
Evan Cheng | Avoid compiler warning: comparison between signed and...
|
commit | commitdiff | tree |
| 2010-10-08 |
Evan Cheng | Don't waste time unfolding simple loads. The unfolded...
|
commit | commitdiff | tree |
| 2010-10-08 |
Evan Cheng | Fix operand latency computation in cases where the...
|
commit | commitdiff | tree |
| 2010-10-07 |
Evan Cheng | Code refactoring.
|
commit | commitdiff | tree |
| 2010-10-07 |
Evan Cheng | Correctly check if a path is a directory. Fix by Brian...
|
commit | commitdiff | tree |
| 2010-10-07 |
Evan Cheng | Canonicalize X86ISD::MOVDDUP nodes to v2f64 to make...
|
commit | commitdiff | tree |
| 2010-10-07 |
Evan Cheng | Model operand cycles of vldm / vstm; also fixes scheduling...
|
commit | commitdiff | tree |
| 2010-10-06 |
Evan Cheng | - Add TargetInstrInfo::getOperandLatency() to compute...
|
commit | commitdiff | tree |
| 2010-10-04 |
Evan Cheng | This DAG combine BRCOND transformation can look pass...
|
commit | commitdiff | tree |
| 2010-10-03 |
Evan Cheng | Major changes to Cortex-A9 itinerary.
|
commit | commitdiff | tree |
| 2010-10-01 |
Evan Cheng | Fix r115332: correctly model AGU / NEON mux.
|
commit | commitdiff | tree |
| 2010-10-01 |
Evan Cheng | Fix scheduling infor for vmovn and vshrn which I broke...
|
commit | commitdiff | tree |
| 2010-10-01 |
Evan Cheng | Add operand cycles for vldr / vstr.
|
commit | commitdiff | tree |
| 2010-10-01 |
Evan Cheng | NEON scheduling info fix. vmov reg, reg are single...
|
commit | commitdiff | tree |
| 2010-10-01 |
Evan Cheng | Per Cortex-A9 pipeline diagram. AGU (core load / store...
|
commit | commitdiff | tree |
| 2010-09-30 |
Evan Cheng | Comments about operand cycles and pipeline forwarding...
|
commit | commitdiff | tree |
| 2010-09-30 |
Evan Cheng | ARM instruction itinerary fixes:
|
commit | commitdiff | tree |
| 2010-09-29 |
Evan Cheng | Model Cortex-a9 load to SUB, RSB, ADD, ADC, SBC, RSC...
|
commit | commitdiff | tree |
| 2010-09-29 |
Evan Cheng | Separate itinerary classes for mvn from mov; for tst...
|
commit | commitdiff | tree |
| 2010-09-29 |
Evan Cheng | Assign bitwise binary instructions different itinerary...
|
commit | commitdiff | tree |
| 2010-09-28 |
Evan Cheng | Add support to model pipeline bypass / forwarding.
|
commit | commitdiff | tree |
| 2010-09-25 |
Evan Cheng | Fix IIC_iEXTAr itinerary class of Cortex-A9.
|
commit | commitdiff | tree |
| 2010-09-25 |
Evan Cheng | Remove a unused instruction itinerary class.
|
commit | commitdiff | tree |
| 2010-09-25 |
Evan Cheng | Fix zero and sign extension instructions scheduling...
|
commit | commitdiff | tree |
| 2010-09-24 |
Evan Cheng | More pseudo instruction scheduling itinerary fixes.
|
commit | commitdiff | tree |
| 2010-09-24 |
Evan Cheng | Fix scheduling itinerary for pseudo mov immediate instructio...
|
commit | commitdiff | tree |
| 2010-09-24 |
Evan Cheng | Enable code placement optimization pass for ARM.
|
commit | commitdiff | tree |
| 2010-09-24 |
Evan Cheng | Fix a potential null dereference bug.
|
commit | commitdiff | tree |
| 2010-09-23 |
Evan Cheng | Fix r114632. Return if the only terminator is an uncondition...
|
commit | commitdiff | tree |
| 2010-09-23 |
Evan Cheng | Revert 114634 for now since buildbot claim it broke...
|
commit | commitdiff | tree |
| 2010-09-23 |
Evan Cheng | Follow up to r114630. Do not optimize away unconditional...
|
commit | commitdiff | tree |
| 2010-09-23 |
Evan Cheng | Disable codegen prepare critical edge splitting. Machine...
|
commit | commitdiff | tree |
| 2010-09-23 |
Evan Cheng | If there are multiple unconditional branches terminating...
|
commit | commitdiff | tree |
| 2010-09-23 |
Evan Cheng | Don't sink insert_subreg, subreg_to_reg, reg_sequence...
|
commit | commitdiff | tree |
| 2010-09-23 |
Evan Cheng | SDISel should not optimize a unconditional branch following...
|
commit | commitdiff | tree |
| 2010-09-21 |
Evan Cheng | OptimizeCompareInstr should avoid iterating pass the...
|
commit | commitdiff | tree |
| 2010-09-20 |
Evan Cheng | Enable machine sinking critical edge splitting. e.g.
|
commit | commitdiff | tree |
| 2010-09-20 |
Evan Cheng | Avoid splitting critical edge twice for a set of PHI...
|
commit | commitdiff | tree |
| 2010-09-18 |
Evan Cheng | Fix code that break critical edges for PHI uses. Watch...
|
commit | commitdiff | tree |
| 2010-09-17 |
Evan Cheng | Teach machine sink to
|
commit | commitdiff | tree |
| 2010-09-17 |
Evan Cheng | Machine CSE was forgetting to clear some data structures.
|
commit | commitdiff | tree |
| 2010-09-17 |
Evan Cheng | Fix a potential bug that can cause miscomparison with...
|
commit | commitdiff | tree |
| 2010-09-10 |
Evan Cheng | Fix test so it passes on non-Darwin hosts.
|
commit | commitdiff | tree |
| 2010-09-10 |
Evan Cheng | Teach if-converter to be more careful with predicating...
|
commit | commitdiff | tree |
| 2010-09-09 |
Evan Cheng | For each instruction itinerary class, specify the number...
|
commit | commitdiff | tree |
| 2010-09-08 |
Evan Cheng | Fix LDM_RET schedule itinery.
|
commit | commitdiff | tree |
| 2010-09-07 |
Evan Cheng | Remove a dead comment.
|
commit | commitdiff | tree |
| 2010-08-19 |
Evan Cheng | Update debug logs.
|
commit | commitdiff | tree |
| 2010-08-19 |
Evan Cheng | Properly update MachineDominators when splitting critical...
|
commit | commitdiff | tree |
| 2010-08-19 |
Evan Cheng | It's possible to sink a def if its local uses are PHI's.
|
commit | commitdiff | tree |
| 2010-08-19 |
Evan Cheng | Remove disabled assertion.
|
commit | commitdiff | tree |
| 2010-08-19 |
Evan Cheng | Teach machine-sink to break critical edges when appropriate...
|
commit | commitdiff | tree |
| 2010-08-18 |
Evan Cheng | If any def of a machine-sink candidate has local uses...
|
commit | commitdiff | tree |
| 2010-08-17 |
Evan Cheng | Add nounwind.
|
commit | commitdiff | tree |
| 2010-08-17 |
Evan Cheng | PHI elimination shouldn't require machineloopinfo since...
|
commit | commitdiff | tree |
| 2010-08-17 |
Evan Cheng | Machine CSE preserves CFG. Pass manager was freeing...
|
commit | commitdiff | tree |
| 2010-08-17 |
Evan Cheng | Move the decision logic whether it's a good idea to...
|
commit | commitdiff | tree |
| 2010-08-17 |
Evan Cheng | Fix debug message.
|
commit | commitdiff | tree |
| 2010-08-17 |
Evan Cheng | Add an option to disable codegen prepare critical edge...
|
commit | commitdiff | tree |
| 2010-08-17 |
Evan Cheng | PHI elimination should not break back edge. It can...
|
commit | commitdiff | tree |
| 2010-08-12 |
Evan Cheng | Make sure ARM constant island pass does not break up...
|
commit | commitdiff | tree |
| 2010-08-11 |
Evan Cheng | Fix test and re-enable it.
|
commit | commitdiff | tree |
| 2010-08-11 |
Evan Cheng | Report error if codegen tries to instantiate a ARM...
|
commit | commitdiff | tree |
| 2010-08-11 |
Evan Cheng | ArchV7M implies HW division instructions.
|
commit | commitdiff | tree |
| 2010-08-11 |
Evan Cheng | ArchV6T2, V7A, and V7M implies Thumb2; Archv7A implies...
|
commit | commitdiff | tree |
| 2010-08-11 |
Evan Cheng | Add ARM Archv6M and let it implies FeatureDB (having...
|
commit | commitdiff | tree |
| next |
projects / oota-llvm.git / search