summaryrefslogtreecommitdiff
AgeCommit message (Expand)AuthorFilesLines
2012-10-18remove unused variable to fix a warning.Nadav Rotem1-2/+1
2012-10-18Add a small example which shows a vectorizable loop with a non-pow-of-two countNadav Rotem1-2/+10
2012-10-18Temporarily revert the TargetTransform changes.Bob Wilson44-821/+317
2012-10-18Remove the use of dominators and AA.Nadav Rotem1-8/+0
2012-10-18Vectorizer: Add support for loops with an unknown count. For example:Nadav Rotem3-73/+162
2012-10-17Revert r166157 because some tests fail...Bill Wendling1-3/+1
2012-10-17Check that the operand of the GEP is not the GEP itself. This occurred during...Bill Wendling1-1/+3
2012-10-17Revert part of r166049 back and enable test case in r166125.Michael Liao2-1/+42
2012-10-17LoopVectorize.cpp: Fix a warning. [-Wunused-variable]NAKAMURA Takumi1-2/+1
2012-10-17Disable extract-concat test case temporarilyMichael Liao1-1/+2
2012-10-17Remove redundant SetInsertPoint call.Jakub Staszak1-1/+0
2012-10-17Revert r166049Michael Liao2-64/+0
2012-10-17Add conditional branch instructions and their patterns.Reed Kotler12-2/+577
2012-10-17Fix some typos and wrong indenting.Roman Divacky1-4/+4
2012-10-17Teach DAG combine to fold (extract_subvec (concat v1, ..) i) to v_iMichael Liao2-2/+34
2012-10-17Switch MRI::UsedPhysRegs to a register unit bit vector.Jakob Stoklund Olesen2-12/+20
2012-10-17Update the release notes about how to enable the loop vectorizer.Nadav Rotem1-2/+2
2012-10-17Add a really faster pre-RA scheduler (-pre-RA-sched=linearize). It doesn't useEvan Cheng4-3/+165
2012-10-17Merge MRI::isPhysRegOrOverlapUsed() into isPhysRegUsed().Jakob Stoklund Olesen4-11/+10
2012-10-17Update the release notes about the store-merge dag optimization.Nadav Rotem1-0/+2
2012-10-17Update the release notes about the new TargetTransformInfo API changes.Nadav Rotem1-0/+5
2012-10-17Update the release notes about the new loop vectorizer.Nadav Rotem1-0/+4
2012-10-17Add a loop vectorizer.Nadav Rotem10-2/+1512
2012-10-17Check for empty YMM use-def lists in X86VZeroUpper.Jakob Stoklund Olesen1-1/+1
2012-10-17Fix fallout from RegInfo => FrameLowering refactoring on MSP430.Anton Korobeynikov5-16/+32
2012-10-17misched: Better handling of invalid latencies in the machine modelAndrew Trick2-5/+15
2012-10-17docs: Add link to integrated assembler HowToSean Silva1-0/+7
2012-10-17Support: Don't remove special files on signals.Daniel Dunbar1-9/+21
2012-10-17[asan] better debug diagnostics in asan compiler moduleKostya Serebryany1-4/+4
2012-10-17This just in, it is a *bad idea* to use 'udiv' on an offset ofChandler Carruth2-3/+23
2012-10-17Marked this variable as 'used' so that LTO doesn't get rid of it.Bill Wendling1-1/+1
2012-10-17Fix a really annoying "bug" introduced in r165941. The change from thatChandler Carruth1-3/+3
2012-10-17Remove LLVM_DELETED_FUNCTION from destructors that override non-deleted base ...Craig Topper1-16/+16
2012-10-17Check SSSE3 instead of SSE4.1Michael Liao1-2/+2
2012-10-17Fix setjmp on models with non-Small code model nor non-Static relocation modelMichael Liao3-24/+66
2012-10-17Use a SparseSet instead of a BitVector for UsedInInstr in RAFast.Jakob Stoklund Olesen2-27/+30
2012-10-16Use a typedef to reduce some typing and reformat code accordingly.Eric Christopher1-12/+17
2012-10-16Variable name cleanup.Eric Christopher1-4/+4
2012-10-16Formatting and 80-col.Eric Christopher1-3/+3
2012-10-16Spacing.Eric Christopher1-1/+1
2012-10-16Avoid rematerializing a redef immediately after the old def.Jakob Stoklund Olesen2-0/+30
2012-10-16Revert r166046 "Switch back to the old coalescer for now to fix the 32 bit bit"Jakob Stoklund Olesen3-368/+2
2012-10-16[InstCombine] Teach InstCombine how to handle an obfuscated splat.Michael Gottesman2-2/+16
2012-10-16[ms-inline asm] Add the helper function, isParseringInlineAsm(). To be used i...Chad Rosier2-0/+2
2012-10-16Simplify code. No functionality change.Jakub Staszak1-7/+3
2012-10-16Check .rela instead of ELF64 for the compensation vaue resettingMichael Liao1-2/+1
2012-10-1680-col fixup.Jakub Staszak1-1/+2
2012-10-16Teach DAG combine to fold (trunc (fptoXi x)) to (fptoXi x)Michael Liao2-0/+64
2012-10-16Switch back to the old coalescer for now to fix the 32 bit bitRafael Espindola3-2/+368
2012-10-16Simplify potentially quadratic behavior while erasing elements from std::vector.Jakub Staszak1-7/+1
generated by cgit v1.2.3 (git 2.39.1) at 2024-04-28 10:52:04 +0000