projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 279706e) | patch
Cost model support for lowered math builtins.
authorBenjamin Kramer <benny.kra@googlemail.com>
Thu, 28 Feb 2013 19:09:33 +0000 (19:09 +0000)
committerBenjamin Kramer <benny.kra@googlemail.com>
Thu, 28 Feb 2013 19:09:33 +0000 (19:09 +0000)
commit8611d4449a77ca05e808823bc966573a85da00cb
tree948b8110ba823e8880a6445c17aaae4dd370d151tree | snapshot
parent279706e90e12e9418d4e8f9415d5f3ed33a99bdbcommit | diff
Cost model support for lowered math builtins.

We make the cost for calling libm functions extremely high as emitting the
calls is expensive and causes spills (on x86) so performance suffers. We still
vectorize important calls like ceilf and friends on SSE4.1. and fabs.

Differential Revision: http://llvm-reviews.chandlerc.com/D466

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@176287 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Analysis/CostModel.cpp diff | blob | history
lib/CodeGen/BasicTargetTransformInfo.cpp diff | blob | history
test/Analysis/CostModel/X86/intrinsic-cost.ll [new file with mode: 0644] blob
C/C++ LLVM-based compilers that forbids OOTA behaviors