- e38993f Test commit. by Tim Northover · 13 years ago
- 1203f2f Enable detection of AVX and AVX2 support through CPUID. Add AVX/AVX2 to corei7-avx, core-avx-i, and core-avx2 cpu names. by Craig Topper · 13 years ago
- 464bda3 Teach the reassociate pass to fold chains of multiplies with repeated by Chandler Carruth · 13 years ago
- cac31de Specify cpu to unbreak tests. by Evan Cheng · 13 years ago
- e67a416 If triple is armv7 / thumbv7 and a CPU is specified, do not automatically assume by Evan Cheng · 13 years ago
- 4866363 Don't forget to reset 'first operand' flag when we're setting the MDNodeOperand value. by Bill Wendling · 13 years ago
- 6962106 Try to fix llvm-arm-linux builder with -mcpu. by Jakob Stoklund Olesen · 13 years ago
- 01c0dd1 Trivial change to make the test use -mcpu=generic so as to avoid by Preston Gurd · 13 years ago
- b856d55 Reapply the SmallMap patch with a fix. by Benjamin Kramer · 13 years ago
- 165324c Print IV chain numbers while collecting them. by Jakob Stoklund Olesen · 13 years ago
- a0b0219 Remove more dead code. by Jakob Stoklund Olesen · 13 years ago
- b56e411 Unify internal representation of ARM instructions with a register right-shifted by #32. These are stored as shifts by #0 in the MCInst and correctly marshalled when transforming from or to assembly representation. by Richard Barton · 13 years ago
- bdbf015 Revert "First implementation of:" by Eric Christopher · 13 years ago
- 76271a3 First implementation of: by Stepan Dyatkovskiy · 13 years ago
- 50e1d84 Simplify LiveIntervals::getApproximateInstructionCount(). by Jakob Stoklund Olesen · 13 years ago
- a62efd8 Remove a dead function. by Jakob Stoklund Olesen · 13 years ago
- 40a2b65 Remove the -disable-cross-class-join option. by Jakob Stoklund Olesen · 13 years ago
- a2404e3 Cross-class joining is winning. by Jakob Stoklund Olesen · 13 years ago
- 8030e1a Add ifdef around getSubtargetFeatureName in tablegen output file so that only targets that want the function get it. This prevents other targets from getting an unused function warning. by Craig Topper · 13 years ago
- c16f851 Use vector_shuffles instead of target specific unpack nodes for AVX ZERO_EXTEND/ANY_EXTEND combine. These will be converted to target specific nodes during lowering. This is more consistent with other code. by Craig Topper · 13 years ago
- fa1059f openbsd doesn't support soname, patch by Brad Smith! by Chris Lattner · 13 years ago
- 3ef91f5 Actually delete now-empty file. by Chandler Carruth · 13 years ago
- 87aac6a Reverting r155468. Chris and Chandler have convinced me that it's dangerous and by Lang Hames · 13 years ago
- 25052f4 Do not use $gp as a dedicated global register if the target ABI is not O32. by Akira Hatanaka · 13 years ago
- 23d59c2 typo in declaration from earlier today by Andrew Trick · 13 years ago
- 50ade65 Simplify the known retain count tracking; use a boolean state instead by Dan Gohman · 13 years ago
- eeeb775 Build custom predecessor and successor lists for each basic block. by Dan Gohman · 13 years ago
- 14ce6fa ARM: improved assembler diagnostics for missing CPU features. by Jim Grosbach · 13 years ago
- 86b7e2a Fix a naughty header include that breaks "installed" builds. by Andrew Trick · 13 years ago
- 80c1ea6 ConstantFoldSelectInstruction swapped the operands of the select. by Nadav Rotem · 13 years ago
- 2003e03 Fix the testcase. We do expect two vblendw on XMMs. by Nadav Rotem · 13 years ago
- 34a13bb Add a testcase for 155440 by Nadav Rotem · 13 years ago
- ddb1420 MachineBasicBlock::SplitCriticalEdge() should follow LLVM IR variant and refuse to break edge to EH landing pad. rdar://11300144 by Evan Cheng · 13 years ago
- 1d9e68d Add support for llvm.arm.neon.vmull* intrinsics to InstCombine. This fixes by Lang Hames · 13 years ago
- 7362ac7 Fix a crash on valid (if UB) bitcode that is produced for some global by Chandler Carruth · 13 years ago
- 95a7e80 ARM: Nuke remnant bogus code. by Jim Grosbach · 13 years ago
- 6ce1c88 Related to PR1255. Let's begin. I'll commit classes that corresponds to our latest PR1255 discussion posts in llvm-commits. by Stepan Dyatkovskiy · 13 years ago
- 7bc9698 AVX: Add additional vbroadcast replacement sequences for integers. by Nadav Rotem · 13 years ago
- 63bbe14 cmake: new file by Andrew Trick · 13 years ago
- c3ad885 misched: DAG builder must special case earlyclobber by Andrew Trick · 13 years ago
- 000b250 misched: try (not too hard) to place debug values where they belong by Andrew Trick · 13 years ago
- eb45ebb misched: ignore debug values during scheduling by Andrew Trick · 13 years ago
- 006e1ab misched: DAG builder support for tracking register pressure within the current scheduling region. by Andrew Trick · 13 years ago
- 4dfeef1 RegisterPressure: A utility for computing register pressure within a by Andrew Trick · 13 years ago
- 24e767d Add missing test cases for ARM VLD3 (single 3-element structure to all lanes) by Kevin Enderby · 13 years ago
- 2c66edf Add missing test cases for ARM VLD4 (single 4-element structure to all lanes) by Kevin Enderby · 13 years ago
- 87ffdbc AVX2: The BLENDPW instruction selects between vectors of v16i16 using an i8 by Nadav Rotem · 13 years ago
- f4478f9 Refactor Thumb ITState handling in ARM Disassembler to more efficiently use its vector by Richard Barton · 13 years ago
- d1a7913 AVX: We lower VECTOR_SHUFFLE and BUILD_VECTOR nodes into vbroadcast instructions by Nadav Rotem · 13 years ago
- adb082c Look for the 'Is Simulated' module flag. This indicates that the program is compiled to run on a simulator. by Bill Wendling · 13 years ago
- 75920ad FileCheck-ize tests. by Bill Wendling · 13 years ago
- c6490d1 FileCheck-ize these tests. by Bill Wendling · 13 years ago
- d5cc8b8 FileCheck-ize these tests. Harden some of them. by Bill Wendling · 13 years ago
- 3ef43cf Remove dangling spaces. Fix some other formatting. by Craig Topper · 13 years ago
- 7fd5e16 Simplify code a bit and make it compile better. Remove unused parameters. by Craig Topper · 13 years ago
- aff5968 Add a missing cpu subtype. by Evan Cheng · 13 years ago
- c92ba4e Tidy up. 80 columns, whitespace, et. al. by Jim Grosbach · 13 years ago
- a354077 Optimize the vector UINT_TO_FP, SINT_TO_FP and FP_TO_SINT operations where the integer type is i8 (commonly used in graphics). by Nadav Rotem · 13 years ago
- 6a8c7bf This patch fixes a problem which arose when using the Post-RA scheduler by Preston Gurd · 13 years ago
- 1d52184 ARM: VSLI two-operand assmebly aliases are tblgen'erated. by Jim Grosbach · 13 years ago
- e1d866e ARM: tblgen'erate VSRA/VRSRA/VSRI assembly two-operand aliases. by Jim Grosbach · 13 years ago
- c34954d ARM: Add testcases for two-operand variants of VSRA/VRSRA/VSRI. by Jim Grosbach · 13 years ago
- 10a3933 Add ARM mode tests for the NEON vector shift-accumulate tests. by Jim Grosbach · 13 years ago
- 2b85250 Tidy up. Reformat for ease of reading. by Jim Grosbach · 13 years ago
- 13b7352 ARM: vqdmulh two-operand aliases are tblgen'erated now. by Jim Grosbach · 13 years ago
- 8dab504 [Support/Unix] Unconditionally include time.h. by Michael J. Spencer · 13 years ago
- 216432d Allow forward declarations to take a context. This helps the debugger by Eric Christopher · 13 years ago
- e3fd2a3 Temporarily revert r155364 until the upstream review can complete, per by Chandler Carruth · 13 years ago
- d410eab Revert r155365, r155366, and r155367. All three of these have regression by Chandler Carruth · 13 years ago
- 15e56ad Hexagon V5 (floating point) support. by Sirish Pande · 13 years ago
- 1bfd248 Support for Hexagon architectural feature, new value jump. by Sirish Pande · 13 years ago
- 0dac391 Support for Hexagon VLIW Packetizer. by Sirish Pande · 13 years ago
- 9f6852d Hexagon Packetizer's target independent fix. by Sirish Pande · 13 years ago
- 72847f3 Reapply r155136 after fixing PR12599. by Jakob Stoklund Olesen · 13 years ago
- 9dc06bd Conflict with st_dev/st_ino identifiers under Debian GNU/Hurd by Sylvestre Ledru · 13 years ago
- 55cabae Fix issue 67 by checking that the interface functions weren't redefined in the compiled source file. by Alexander Potapenko · 13 years ago
- 2d5fdf8 [tsan] use llvm/ADT/Statistic.h for tsan stats by Kostya Serebryany · 13 years ago
- 708e44f Use MVT instead of EVT through all of LowerVECTOR_SHUFFLEtoBlend and not just the switch. Saves a little bit of binary size. by Craig Topper · 13 years ago
- 9d35240 Make getZeroVector and getOnesVector more alike as far as how they detect 128-bit versus 256-bit vectors. Be explicit about both sizes and use llvm_unreachable. Similar changes to getLegalSplat. by Craig Topper · 13 years ago
- 69947b9 Tidy up by removing some 'else' after 'return' by Craig Topper · 13 years ago
- 1842ba0 Tidy up spacing in LowerVECTOR_SHUFFLEtoBlend. Remove code that checks if shuffle operand has a different type than the the shuffle result since it can never happen. by Craig Topper · 13 years ago
- 731dfd0 Add a couple llvm_unreachables. by Craig Topper · 13 years ago
- 0fbf364 Remove some tab characers. by Craig Topper · 13 years ago
- e8eb116 Remove some 'else' after 'return'. No functional change. by Craig Topper · 13 years ago
- a9963c6 Don't die with an assertion if the Result bitwidth is already correct. This by Chris Lattner · 13 years ago
- 85d043d Cleanup whitespace. by Bill Wendling · 13 years ago
- 098c595 Limit the number of times we recurse through this algorithm. All of the by Bill Wendling · 13 years ago
- b14940a Make Extract128BitVector and Insert128BitVector take an unsigned instead of an ConstantNode SDValue. getConstant was almost always called just before only to have the functions take it apart and build a new ConstantSDNode. by Craig Topper · 13 years ago
- 767b4f6 Convert getNode(UNDEF) to getUNDEF. by Craig Topper · 13 years ago
- df966f6 Make calls to getVectorShuffle more consistent. Use shuffle VT for calls to getUNDEF instead of requerying. Use &Mask[0] instead of Mask.data(). by Craig Topper · 13 years ago
- d63fa65 Tidy up. 80 columns and argument alignment. by Craig Topper · 13 years ago
- 4c7972d Simplify code by converting multiple places that were manually concatenating 128-bit vectors to use either CONCAT_VECTORS or a helper function. CONCAT_VECTORS will itself be lowered to the same pattern as before. The helper function is needed for concats of BUILD_VECTORs since getNode(CONCAT_VECTORS) will just return a large BUILD_VECTOR and we may be trying to lower large BUILD_VECTORS when this occurs. by Craig Topper · 13 years ago
- dd90478 cleaned line endings in the newly added test file by Elena Demikhovsky · 13 years ago
- bfae1fd ARM: Initialize the HasRAS bit. by Benjamin Kramer · 13 years ago
- a3e3481 Tidy up this test more: by Chandler Carruth · 13 years ago
- 71f8bc3 FileCheck-ize a test, and tidy it up a touch. by Chandler Carruth · 13 years ago
- 1da5867 ZERO_EXTEND/SIGN_EXTEND/TRUNCATE optimization for AVX2 by Elena Demikhovsky · 13 years ago
- eb23f9e Remove some potential warnings about variables used uninitialized. by Bill Wendling · 13 years ago
- d46575f Add a flag to the struct type finder to collect only those types which have by Bill Wendling · 13 years ago
- 2dbd784 No need for "else if" after a return. Autosense "0o123" as octal in by Chris Lattner · 13 years ago