- e1dfc7d Avoid an unnecessary call to allnodes_size(), which is linear. by Dan Gohman · 17 years ago
- 58e39b0 One more permutation of subtracting off a base value. by Dale Johannesen · 17 years ago
- 87c8a8f Added support for vector widening. by Mon P Wang · 17 years ago
- 84aff84 Fix expansion of vsetcc to set the high bit for true instead of 1. by Mon P Wang · 17 years ago
- 89b64bd Double the amount of memory reserved for SUnits. This is a by Dan Gohman · 17 years ago
- 28dc98f Fix for PR3225: disable a broken optimization in by Eli Friedman · 17 years ago
- fd3b7b7 A new dag combine; several permutations of this by Dale Johannesen · 17 years ago
- 905ff1e Preserve SourceValue information when lowering produces multiple loads from by Dan Gohman · 17 years ago
- 09dc9c0 We have decided not to support inline asm where an output operand with a matching input operand with incompatible type (i.e. either one is a floating point and the other is an integer or the sizes of the types differ). SelectionDAGBuild will catch these and exit with an error. by Evan Cheng · 17 years ago
- 9a65d6a Remove some special-case logic in ScheduleDAGSDNodes's by Dan Gohman · 17 years ago
- 3f23744 Fix some register-alias-related bugs in the post-RA scheduler liveness by Dan Gohman · 17 years ago
- ffa3912 Make addPred and removePred return void, since the return value is not by Dan Gohman · 17 years ago
- 93b3b92 Added support for splitting and scalarizing vector shifts. by Mon P Wang · 17 years ago
- 464fc5a Fix printing of PseudoSourceValues in SDNode graphs. by Dan Gohman · 17 years ago
- d17c030 Added support to LegalizeType for expanding the operands of scalar to vector by Mon P Wang · 17 years ago
- b6862bb Reapply r60997, this time without forgetting that by Duncan Sands · 17 years ago
- 2d8cc31 Temporarily revert r60997. It was causing this failure: by Bill Wendling · 17 years ago
- 0f546d4 LegalizeDAG is not supposed to introduce illegal by Duncan Sands · 17 years ago
- c5d6460 Remove assertion to allow promotion of a truncating store operand by Mon P Wang · 17 years ago
- e1a0b2e Added basic support for expanding VSETCC by Mon P Wang · 17 years ago
- fd6673c On big-endian machines it is wrong to do a full by Duncan Sands · 17 years ago
- 05e11fa When expanding unaligned loads and stores do not make by Duncan Sands · 17 years ago
- 637ed03 Fix add/sub expansion: don't create ADD / SUB with two results (seems like everyone is doing this these days :-). Patch by Daniel M Gessel! by Evan Cheng · 17 years ago
- 712f7b3 When using a 4 byte jump table on a 64 bit machine, by Duncan Sands · 17 years ago
- 9e5ecb8 Added support for SELECT v8i8 v4i16 for X86 (MMX) by Mon P Wang · 17 years ago
- ab55ebd Redo the arithmetic with overflow architecture. I was changing the semantics of by Bill Wendling · 17 years ago
- 6db08c4 Avoid generating a convert_rndsat node when the src and dest type are the same. by Mon P Wang · 17 years ago
- 775bb80 Clarify FIXME. by Bill Wendling · 17 years ago
- f67303d Whitespace clean up (tabs with spaces) by Mon P Wang · 17 years ago
- a60b523 Make fix for r60829 less conservative to allow the proper optimization for by Mon P Wang · 17 years ago
- 2476e5d If ADD, SUB, or MUL have an overflow bit that's used, don't do transformation on by Bill Wendling · 17 years ago
- ab0c578 For amusement, implement SADDO, SSUBO, UADDO, USUBO by Duncan Sands · 17 years ago
- bae527d Fixed a bug when trying to optimize a extract vector element of a by Mon P Wang · 17 years ago
- bc5e15e Whitespace changes. by Bill Wendling · 17 years ago
- 2ce4e5c Whitespace fixes. by Bill Wendling · 17 years ago
- 54e4c36 Rewrite the SDep class, and simplify some of the related code. by Dan Gohman · 17 years ago
- 74c3765 Add sub/mul overflow intrinsics. This currently doesn't have a by Bill Wendling · 17 years ago
- 47d9dcc Fix PR3117: not all nodes being legalized. The by Duncan Sands · 17 years ago
- 1cd46bb In LegalizeOp, don't change the result type of CONVERT_RNDSAT when promoting by Mon P Wang · 17 years ago
- e9f1015 Fix getNode to allow a vector for the shift amount for shifts of vectors. by Mon P Wang · 17 years ago
- c8a1a3c Factor out the code for sign-extending/truncating gep indices by Dan Gohman · 17 years ago
- 9304f2c When allocating a stack temporary, use the correct by Duncan Sands · 17 years ago
- b29ffc8 Make debug output more informative. by Dan Gohman · 17 years ago
- 8b8e5a7 Only check that the result of the mapping was not by Duncan Sands · 17 years ago
- 96da041 Remove a (what appears to be) overly strict assertion. Here is what happened: by Evan Cheng · 17 years ago
- 546d7b5 Non-functional change: make custom lowering for truncate stylistically by Scott Michel · 17 years ago
- 34d7985 One more transformation. by Dale Johannesen · 17 years ago
- b0a5cdd make it possible to custom lower TRUNCATE (needed for the CellSPU target) by Tilmann Scheller · 17 years ago
- fb13f00 Removed some unnecessary code in widening. by Mon P Wang · 17 years ago
- 221cd2f Add a few more transformations. by Dale Johannesen · 17 years ago
- 13d6d44 Expand getVTList, getNodeValueTypes, and SelectNodeTo to handle more value types. by Bill Wendling · 17 years ago
- aaffa05 There are no longer any places that require a by Duncan Sands · 17 years ago
- 1607f05 Change the interface to the type legalization method by Duncan Sands · 17 years ago
- fc69cb4 Followup to r60283: optimize arbitrary width signed divisions as well by Eli Friedman · 17 years ago
- 201c977 Fix for PR2164: allow transforming arbitrary-width unsigned divides into by Eli Friedman · 17 years ago
- 86f874d APIntify a test which is potentially unsafe otherwise, and fix the by Eli Friedman · 17 years ago
- 9b0f0b5 Removing redundant semicolons. No functionality change. by Sanjiv Gupta · 17 years ago
- 56eca91 Add a missing case in visitADD. by Dale Johannesen · 17 years ago
- d3f01aa Allow custom lowering of ADDE/ADDC/SUBE/SUBC operations. by Sanjiv Gupta · 17 years ago
- 740464e A simplification for checking whether the signs of the operands and sum differ. Thanks, Duncan. by Bill Wendling · 17 years ago
- a6af91a Now with the correct type for the 0. by Bill Wendling · 17 years ago
- 7ebfea4 Get rid of unused variable. by Bill Wendling · 17 years ago
- c0062fb Hacker's Delight says, "Signed integer overflow of addition occurs if and only by Bill Wendling · 17 years ago
- 21d9003 Initial support for anti-dependence breaking. Currently this code does not by Dan Gohman · 17 years ago
- 41ea7e7 - Make lowering of "add with overflow" customizable by back-ends. by Bill Wendling · 17 years ago
- 942b8c8 Check in the rest of this change. The isAntiDep flag needs to be passed by Dan Gohman · 17 years ago
- 25cf227 If the type legalizer actually legalized anything by Duncan Sands · 17 years ago
- 24ac408 Eliminate some unused variable compile time warnings. by Evan Cheng · 17 years ago
- def2739 Add support for llvm.uadd.with.overflow. by Bill Wendling · 17 years ago
- 0322808 Rename SetCCResultContents to BooleanContents. In by Duncan Sands · 17 years ago
- 6cce3da Added check to avoid generating extract subvector beyond the end of the vector when normalizing vector shuffles. by Mon P Wang · 17 years ago
- 253174b Cleanup of the [SU]ADDO type legalization code. Patch by Duncan! by Bill Wendling · 17 years ago
- 8ac0d4b - Move conversion of [SU]ADDO from DAG combiner into legalizer. by Bill Wendling · 17 years ago
- 3ebd0ee Update comments. by Dan Gohman · 17 years ago
- 824b958 reapply Sanjiv's patch to genericize memcpy/memset/memmove to take an by Chris Lattner · 17 years ago
- dc77092 Revert r59802. It was breaking the build of llvm-gcc: by Bill Wendling · 17 years ago
- 219ab40 Make mem[cpy,move,set] intrinsics overloaded. by Sanjiv Gupta · 17 years ago
- 137dccd Default to converting UADDO to the generic form that SADDO is converted to. by Bill Wendling · 17 years ago
- 230e4fa Clean up normalization of shuffles by Mon P Wang · 17 years ago
- ef37546 Combine the two add with overflow intrinsics lowerings. They differ only in DAG node type. by Bill Wendling · 17 years ago
- da4948a Generate code for llvm.uadd.with.overflow intrinsic. No conversion support yet. by Bill Wendling · 17 years ago
- cddd428 Add a flag to SDep for tracking which edges are anti-dependence edges. by Dan Gohman · 17 years ago
- 6c63f62 Remove chains. Unnecessary. by Bill Wendling · 17 years ago
- 98adea1 Rename SDep's isSpecial to isArtificial, to make this field a little by Dan Gohman · 17 years ago
- 1c55a9c Rename "ADDO" to "SADDO" and "UADDO". The "UADDO" isn't equivalent to "ADDC" by Bill Wendling · 17 years ago
- 7cdc3c8 Implement the sadd_with_overflow intrinsic. This is converted into by Bill Wendling · 17 years ago
- 787782f Use ComputeLatency in the MachineInstr scheduler. by Dan Gohman · 17 years ago
- f209c2c Remove the CycleBound computation code from the ScheduleDAGRRList by Dan Gohman · 17 years ago
- c8c2827 Implement ComputeLatency for MachineInstr ScheduleDAGs. Factor by Dan Gohman · 17 years ago
- ea7b527 Add UADDO and SADDO nodes. These will be used for determining an overflow by Bill Wendling · 17 years ago
- 47d1a21 Change these schedulers to not emit no-ops. It turns out that by Dan Gohman · 17 years ago
- 810260b Add some documentation. by Duncan Sands · 17 years ago
- f0a2d0c 80-column violation. by Bill Wendling · 17 years ago
- 57bebbb Remove a remnant of list-burr's fast mode. by Dan Gohman · 17 years ago
- 117f3e9 Factor out the SethiUllman numbering logic from the list-burr and by Dan Gohman · 17 years ago
- 9e76fea Remove the "fast" form of the list-burr scheduler, and use the by Dan Gohman · 17 years ago
- 6be2ee4 Facter AddPseudoTwoAddrDeps and associated infrasructure out of by Dan Gohman · 17 years ago
- a1e6d36 Factor out the code for verifying the work of the scheduler, by Dan Gohman · 17 years ago
- cdb260d Simplify this code a little. In the fast scheduler, CreateNewSUnit by Dan Gohman · 17 years ago
- 343f0c0 Experimental post-pass scheduling support. Post-pass scheduling by Dan Gohman · 17 years ago