Sanjay Patel d26914da05 [x86, AVX] replace masked load with full vector load when possible ...

Converting masked vector loads to regular vector loads for x86 AVX should always be a win.
I raised the legality issue of reading the extra memory bytes on llvm-dev. I did not see any
objections.

1. x86 already does this kind of optimization for multiple scalar loads -> vector load.
2. If other targets have the same flexibility, we could move this transform up to CGP or DAGCombiner.

Differential Revision: http://reviews.llvm.org/D18094



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@263446 91177308-0d34-0410-b5e6-96231b3b80d8

2016-03-14 16:54:43 +00:00

..

Analysis

ConstantFoldInstruction: avoid wasted calls to ConstantFoldConstantExpression

2016-03-13 05:36:15 +00:00

AsmParser

Add a flag to the LLVMContext to disable name for Value other than GlobalValue

2016-03-10 01:28:54 +00:00

Bitcode

[Bitcode] Make writeComdats less strange

2016-03-13 08:01:03 +00:00

CodeGen

Revert "Recommitted r261633 "Supporting all entities declared in lexical scope in LLVM debug info." After fixing PR26715 at r263379."

2016-03-14 14:58:36 +00:00

DebugInfo

[DebugInfo] Dump CIE augmentation data as a list of hex bytes

2016-03-01 18:38:05 +00:00

ExecutionEngine

ExecutionEngine: tweak debug log

2016-03-05 20:00:41 +00:00

Fuzzer

[libFuzzer] try to use max_len based on the items of the corpus instead of blindly defaulting to 64 bytes.

2016-03-12 01:57:04 +00:00

IR

Add facility to add/remove/check attribute on function and arguments.

2016-03-14 01:37:29 +00:00

IRReader

…

LibDriver

…

LineEditor

…

Linker

Minor cleanup and documentation to IRMover (NFC)

2016-03-11 22:19:06 +00:00

LTO

[ThinLTO] Support for reference graph in per-module and combined summary.

2016-03-11 18:52:24 +00:00

MC

Don't crash when compiling inline assembler containing .file directives.

2016-03-09 17:32:56 +00:00

Object

[ThinLTO] Support for reference graph in per-module and combined summary.

2016-03-11 18:52:24 +00:00

ObjectYAML

[AMDGPU] add AMDGPU target support to ELFObjectFile.h header

2016-03-09 17:08:19 +00:00

Option

…

Passes

[PM] Rename the CRTP mixin base classes for the new pass manager to

2016-03-11 10:33:22 +00:00

ProfileData

Use LineLocation instead of CallsiteLocation to index callsite profile.

2016-03-03 18:09:32 +00:00

Support

APFloat: Fix ilogb for denormals

2016-03-13 05:12:32 +00:00

TableGen

[TableGen] Fix typos in comments. NFC

2016-02-26 06:50:27 +00:00

Target

[x86, AVX] replace masked load with full vector load when possible

2016-03-14 16:54:43 +00:00

Transforms

[CVP] Replace nonnegative with positive, per Philip's request. NFC.

2016-03-14 13:48:00 +00:00

CMakeLists.txt

Move ObjectYAML code to a new library.

2016-03-01 19:15:06 +00:00

LLVMBuild.txt

Add LLVMBuild for ObjectYAML.

2016-03-01 21:29:33 +00:00