[PATCH] Vectorize math intrinsic calls in SLPVectorizer

[Karthik Bhat]

Thanks Tyler, Nadav. Committed as r207901.
I have run llvm lit regressions using make check-all . I'm a bit new here not sure how to execute performance test suites for llvm. If you could guide me to some documentation i would be happy to test and see if this patch improves any workload in llvm performance test cases.
Thanks
Karthik

http://reviews.llvm.org/D3560