[PATCH] D14829: [SLP] Vectorize gather-like idioms ending at non-consecutive loads.

[Matthew Simpson via llvm-commits]

mssimpso added a comment.

Hi All,

I've updated this patch to eliminate the previously observed compile-time regression in MiBench/security-rijndael. I fixed the issue by moving the contiguous access check (the most expensive check prior to vectorization) to the last step and by processing the accesses in chunks of 16. These two changes follow the existing flow we have for stores to minimze compile-time.

I benchmarked this change together with two others. The additional changes include the type-shrinking work I previously mentioned (http://reviews.llvm.org/D15815) and an additional cost model hook (http://reviews.llvm.org/D15816) to catch the sign extensions we introduce with the type-shrinking. Feel free to comment on those patches as well if interested. Together, the patch set causes no compile-time regressions and improves spec2006/h264ref by ~6% on our Cortex-A57-like architecture.

Thanks!


http://reviews.llvm.org/D14829