Linux build

continuous-linux-master

X86: avoid vector-scalar shifts if splat amount is directly a vector …

Windows build

X86: disable K-masks for AVX512BW+VL

Their usage often generates code that is ineffective on SKX.
Use conservative approach for xmm/ymm byte/word vectors.

Continuous build (linux-release_60)

Travis CI build log: https://travis-ci.org/RPCS3/llvm/builds/394240930/

Continuous build (linux-continuous-master)

Travis CI build log: https://travis-ci.org/RPCS3/llvm/builds/394318203/

Continuous build (linux-continuous-linux-master)

Travis CI build log: https://travis-ci.org/RPCS3/llvm/builds/394311891/

Continuous build (linux-continuous-linux-continuous-linux-master)

Travis CI build log: https://travis-ci.org/RPCS3/llvm/builds/394314038/

Continuous build (linux-continuous-linux-continuous-linux-continuous-linux-master)

Travis CI build log: https://travis-ci.org/RPCS3/llvm/builds/394315416/

Continuous build (linux-continuous-linux-continuous-linux-continuous-linux-continuous-linux-master)

Travis CI build log: https://travis-ci.org/RPCS3/llvm/builds/394316206/

continuous-release_60

Prebuilt LLVM Libraries

continuous-release_60

[ValueLattice] Use union to shave off ptr size bytes from elements.

By using a union for Constant* and ConstantRange we can shave off ptr
size bytes off lattice elements. On 64 bit systems, it brings down the
size to 40 bytes from 48 bytes.

Initialization of Range happens on-demand using placement new, if the
state changes to constantrange from non-constantrange. Similarly, the
Range object is destroyed if the state changes from constantrange to
non-constantrange.

Reviewers: reames, anna, davide

Reviewed By: reames, davide

Differential Revision: https://reviews.llvm.org/D41903


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@323049 91177308-0d34-0410-b5e6-96231b3b80d8