2008-11-25 |
Bill Wendling | A simplification for checking whether the signs of... |
tree | commitdiff |
2008-11-25 |
Bill Wendling | Now with the correct type for the 0. |
tree | commitdiff |
2008-11-25 |
Bill Wendling | Get rid of unused variable. |
tree | commitdiff |
2008-11-25 |
Bill Wendling | Hacker's Delight says, "Signed integer overflow of... |
tree | commitdiff |
2008-11-25 |
Dan Gohman | Initial support for anti-dependence breaking. Currently... |
tree | commitdiff |
2008-11-24 |
Bill Wendling | - Make lowering of "add with overflow" customizable... |
tree | commitdiff |
2008-11-24 |
Dan Gohman | Check in the rest of this change. The isAntiDep flag... |
tree | commitdiff |
2008-11-24 |
Duncan Sands | If the type legalizer actually legalized anything |
tree | commitdiff |
2008-11-24 |
Evan Cheng | Eliminate some unused variable compile time warnings. |
tree | commitdiff |
2008-11-24 |
Bill Wendling | Add support for llvm.uadd.with.overflow. |
tree | commitdiff |
2008-11-23 |
Duncan Sands | Rename SetCCResultContents to BooleanContents. In |
tree | commitdiff |
2008-11-23 |
Mon P Wang | Added check to avoid generating extract subvector beyon... |
tree | commitdiff |
2008-11-22 |
Bill Wendling | Cleanup of the [SU]ADDO type legalization code. Patch... |
tree | commitdiff |
2008-11-22 |
Bill Wendling | - Move conversion of [SU]ADDO from DAG combiner into... |
tree | commitdiff |
2008-11-21 |
Dan Gohman | Update comments. |
tree | commitdiff |
2008-11-21 |
Chris Lattner | reapply Sanjiv's patch to genericize memcpy/memset... |
tree | commitdiff |
2008-11-21 |
Bill Wendling | Revert r59802. It was breaking the build of llvm-gcc: |
tree | commitdiff |
2008-11-21 |
Sanjiv Gupta | Make mem[cpy,move,set] intrinsics overloaded. |
tree | commitdiff |
2008-11-21 |
Bill Wendling | Default to converting UADDO to the generic form that... |
tree | commitdiff |
2008-11-21 |
Mon P Wang | Clean up normalization of shuffles |
tree | commitdiff |
2008-11-21 |
Bill Wendling | Combine the two add with overflow intrinsics lowerings... |
tree | commitdiff |
2008-11-21 |
Bill Wendling | Generate code for llvm.uadd.with.overflow intrinsic... |
tree | commitdiff |
2008-11-21 |
Dan Gohman | Add a flag to SDep for tracking which edges are anti... |
tree | commitdiff |
2008-11-21 |
Bill Wendling | Remove chains. Unnecessary. |
tree | commitdiff |
2008-11-21 |
Dan Gohman | Rename SDep's isSpecial to isArtificial, to make this... |
tree | commitdiff |
2008-11-21 |
Bill Wendling | Rename "ADDO" to "SADDO" and "UADDO". The "UADDO" isn... |
tree | commitdiff |
2008-11-21 |
Bill Wendling | Implement the sadd_with_overflow intrinsic. This is... |
tree | commitdiff |
2008-11-21 |
Dan Gohman | Use ComputeLatency in the MachineInstr scheduler. |
tree | commitdiff |
2008-11-21 |
Dan Gohman | Remove the CycleBound computation code from the Schedul... |
tree | commitdiff |
2008-11-21 |
Dan Gohman | Implement ComputeLatency for MachineInstr ScheduleDAGs... |
tree | commitdiff |
2008-11-21 |
Bill Wendling | Add UADDO and SADDO nodes. These will be used for deter... |
tree | commitdiff |
2008-11-21 |
Dan Gohman | Change these schedulers to not emit no-ops. It turns... |
tree | commitdiff |
2008-11-20 |
Duncan Sands | Add some documentation. |
tree | commitdiff |
2008-11-20 |
Bill Wendling | 80-column violation. |
tree | commitdiff |
2008-11-20 |
Dan Gohman | Remove a remnant of list-burr's fast mode. |
tree | commitdiff |
2008-11-20 |
Dan Gohman | Factor out the SethiUllman numbering logic from the... |
tree | commitdiff |
2008-11-20 |
Dan Gohman | Remove the "fast" form of the list-burr scheduler,... |
tree | commitdiff |
2008-11-20 |
Dan Gohman | Facter AddPseudoTwoAddrDeps and associated infrasructur... |
tree | commitdiff |
2008-11-20 |
Dan Gohman | Factor out the code for verifying the work of the sched... |
tree | commitdiff |
2008-11-19 |
Dan Gohman | Simplify this code a little. In the fast scheduler... |
tree | commitdiff |
2008-11-19 |
Dan Gohman | Experimental post-pass scheduling support. Post-pass... |
tree | commitdiff |
2008-11-19 |
Dan Gohman | Move the code for printing a graph node label for an... |
tree | commitdiff |
2008-11-19 |
Dan Gohman | Convert SUnit's dump method into a print method and... |
tree | commitdiff |
2008-11-19 |
Dan Gohman | Rearrange code to reduce the nesting level. No function... |
tree | commitdiff |
2008-11-19 |
Dan Gohman | Fix debug printing of flagged SDNodes in SUnits so... |
tree | commitdiff |
2008-11-18 |
Dan Gohman | Tidy up ScheduleNodeBottomUp methods, and make them... |
tree | commitdiff |
2008-11-18 |
Dan Gohman | Update a comment to reflect the current code. |
tree | commitdiff |
2008-11-18 |
Duncan Sands | Remove integer promotion support for FP_EXTEND |
tree | commitdiff |
2008-11-18 |
Duncan Sands | Simplify code using helper routines. There is not |
tree | commitdiff |
2008-11-18 |
Dan Gohman | Whitespace cleanups. |
tree | commitdiff |
2008-11-18 |
Duncan Sands | LegalizeTypes support for splitting and scalarizing |
tree | commitdiff |
2008-11-18 |
Bill Wendling | Rename stackprotector_create intrinsic to stackprotector. |
tree | commitdiff |
2008-11-18 |
Duncan Sands | Reapply r59464, this time using the correct type |
tree | commitdiff |
2008-11-18 |
Bill Wendling | Remove the stackprotector_check intrinsic. Use a volati... |
tree | commitdiff |
2008-11-18 |
Dan Gohman | Fix a typo in a comment. |
tree | commitdiff |
2008-11-18 |
Dan Gohman | Change SUnit's dump method to take a ScheduleDAG* inste... |
tree | commitdiff |
2008-11-18 |
Bill Wendling | Revert r59464. It was causing this failure: |
tree | commitdiff |
2008-11-18 |
Dan Gohman | Avoid using a loop in ReleasePred and ReleaseSucc metho... |
tree | commitdiff |
2008-11-17 |
Dan Gohman | Eliminate some trivial differences between the Schedule... |
tree | commitdiff |
2008-11-17 |
Duncan Sands | Add soft float support for a bunch more operations... |
tree | commitdiff |
2008-11-17 |
Dan Gohman | Don't bother doing latency calculations in the "fast... |
tree | commitdiff |
2008-11-17 |
Dan Gohman | Use SUnit's CycleBound field instead of duplicating... |
tree | commitdiff |
2008-11-17 |
Richard Osborne | Don't produce ADDC/ADDE when expanding SHL unless they... |
tree | commitdiff |
2008-11-17 |
Dan Gohman | Don't use the isPending flag to mean what the isAvailab... |
tree | commitdiff |
2008-11-17 |
Mon P Wang | Fixed legalization of CONVERT_RNDSAT for integers. |
tree | commitdiff |
2008-11-16 |
Mon P Wang | Improved shuffle normalization to avoid using extract... |
tree | commitdiff |
2008-11-15 |
Duncan Sands | When splitting a SHUFFLE_VECTOR, try to have the result |
tree | commitdiff |
2008-11-15 |
Mon P Wang | Add missing widen operations, fixed widening for extrac... |
tree | commitdiff |
2008-11-15 |
Dan Gohman | Correct a comment. |
tree | commitdiff |
2008-11-15 |
Dan Gohman | Move ScheduleDAGList's LatencyPriorityQueue class out... |
tree | commitdiff |
2008-11-14 |
Dan Gohman | Add support for building a ScheduleDAG from MachineInst... |
tree | commitdiff |
2008-11-14 |
Dan Gohman | For post-regalloc scheduling, remove the instructions... |
tree | commitdiff |
2008-11-14 |
Dan Gohman | Check in the correct version of the patch in r59279. |
tree | commitdiff |
2008-11-14 |
Dan Gohman | Debug printing for SUnits that carry MachineInstrs. |
tree | commitdiff |
2008-11-14 |
Dan Gohman | Initial support for carrying MachineInstrs in SUnits. |
tree | commitdiff |
2008-11-13 |
Dan Gohman | Change DOTGraphTraits<ScheduleDAG*>::getGraphName how... |
tree | commitdiff |
2008-11-13 |
Dan Gohman | Remove the FlaggedNodes member from SUnit. Instead... |
tree | commitdiff |
2008-11-13 |
Dan Gohman | Make the Node member of SUnit private, and add accessors. |
tree | commitdiff |
2008-11-13 |
Dan Gohman | Change ScheduleDAG's DAG member from a reference to... |
tree | commitdiff |
2008-11-12 |
Dan Gohman | Move the code that inserts X87 FP_REG_KILL instructions... |
tree | commitdiff |
2008-11-12 |
Dale Johannesen | Fix unsigned char->ppcf128 conversion. |
tree | commitdiff |
2008-11-12 |
Duncan Sands | Simplify SplitVecRes_EXTRACT_SUBVECTOR. This means |
tree | commitdiff |
2008-11-12 |
Duncan Sands | Correct some thinkos in the expansion of ADD/SUB |
tree | commitdiff |
2008-11-12 |
Dale Johannesen | Fix the testb optimization so x86 also bootstraps. |
tree | commitdiff |
2008-11-11 |
Dan Gohman | In ScheduleDAGRRList::CopyAndMoveSuccessors, create... |
tree | commitdiff |
2008-11-11 |
Dan Gohman | Avoid relying on the SelectionDAG for initializing... |
tree | commitdiff |
2008-11-11 |
Dan Gohman | Change the scheduler accessor methods to accept an... |
tree | commitdiff |
2008-11-11 |
Bill Wendling | Fix for PR3040: |
tree | commitdiff |
2008-11-11 |
Mon P Wang | Cleaned up and fix bugs in convert_rndsat node |
tree | commitdiff |
2008-11-10 |
Bill Wendling | Temporarily revert r58979 and related patch. It's causi... |
tree | commitdiff |
2008-11-10 |
Mon P Wang | Added CONVERT_RNDSAT (conversion with rounding and... |
tree | commitdiff |
2008-11-10 |
Duncan Sands | Fix PR2667: add soft float support for sint_to_fp/uint_... |
tree | commitdiff |
2008-11-10 |
Duncan Sands | Tweak some comments. |
tree | commitdiff |
2008-11-10 |
Duncan Sands | Small cleanups. No functionality change intended! |
tree | commitdiff |
2008-11-10 |
Duncan Sands | When promoting the result of fp_to_uint/fp_to_sint, |
tree | commitdiff |
2008-11-10 |
Dale Johannesen | Really fix testb optimization on big-endian. |
tree | commitdiff |
2008-11-10 |
Mon P Wang | Added support for the following definition of shufflevector |
tree | commitdiff |
2008-11-09 |
Dale Johannesen | Temporarily revert 58825, which breaks PPC bootstrap. |
tree | commitdiff |
2008-11-08 |
Duncan Sands | Try to produce better code when scalarizing VSETCC. |
tree | commitdiff |
2008-11-08 |
Dale Johannesen | Make testb optimization work on big-endian targets. |
tree | commitdiff |
next |