emscripten-fastcomp/test/CodeGen/X86/fdiv.ll, branch 1.13.2 LLVM with the emscripten fastcomp javascript backend https://git.amat.us/emscripten-fastcomp/atom/test/CodeGen/X86/fdiv.ll?h=1.13.2 2012-04-10T20:35:27Z Add a comment noting that the fdiv -> fmul conversion won't generate 2012-04-10T20:35:27Z Duncan Sands baldrick@free.fr 2012-04-10T20:35:27Z urn:sha1:507bb7a42fce6502f90268712cbc32158c17bb59 multiplication by a denormal, and some tests checking that. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@154431 91177308-0d34-0410-b5e6-96231b3b80d8 Only have codegen turn fdiv by a constant into fmul by the reciprocal 2012-04-08T18:08:12Z Duncan Sands baldrick@free.fr 2012-04-08T18:08:12Z urn:sha1:3ef3fcfc0414b96d626b706c6151596684b7d134 when -ffast-math, i.e. don't just always do it if the reciprocal can be formed exactly. There is already an IR level transform that does that, and it does it more carefully. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@154296 91177308-0d34-0410-b5e6-96231b3b80d8 Convert floating point division by a constant into multiplication by the 2012-04-07T20:04:00Z Duncan Sands baldrick@free.fr 2012-04-07T20:04:00Z urn:sha1:961d666be4f3714452ff9f15470f9255d85f0506 reciprocal if converting to the reciprocal is exact. Do it even if inexact if -ffast-math. This substantially speeds up ac.f90 from the polyhedron benchmarks. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@154265 91177308-0d34-0410-b5e6-96231b3b80d8