Jingyue Wu eb653bd9d1 [NVPTX] run LSR before straight-line optimizations
Summary:
Straight-line optimizations can simplify the loop body and make LSR's
cost analysis more precise. This significantly improves several Eigen3
CUDA benchmarks.

With this change, EigenContractionKernel runs up to 40% faster
(753ceee5f2/unsupported/Eigen/CXX11/src/Tensor/TensorContractionCuda.h (cl-502)).
EigenConvolutionKernel2D runs up to 10% faster
(753ceee5f2/unsupported/Eigen/CXX11/src/Tensor/TensorConvolution.h (cl-605)).

I have some difficulties writing small tests that benefit from this
reordering due to a seemingly issue with LSR (being discussed at
http://lists.cs.uiuc.edu/pipermail/llvmdev/2015-July/088244.html).

See the review thread for the compilation time impact of GVN. 

Reviewers: eliben, jholewinski

Subscribers: llvm-commits, jholewinski

Differential Revision: http://reviews.llvm.org/D11304

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@242982 91177308-0d34-0410-b5e6-96231b3b80d8
2015-07-23 04:59:07 +00:00
..