[llvm-commits] [llvm] r55823 - in /llvm/trunk/lib/CodeGen/SelectionDAG: SelectionDAGBuild.cpp SelectionDAGBuild.h

Dale Johannesen dalej at apple.com
Fri Sep 5 09:54:41 PDT 2008


On Sep 5, 2008, at 1:04 AMPDT, Duncan Sands wrote:

> Hi Dale,
>
>> Add -flimit-float-precision to enable some faster,
>> but less accurate (non-IEEE) code sequences for
>> certain math library functions.  Add the first of
>> several such expansions.  Don't worry, if you don't
>> turn it on it won't affect you.
>
> Why not use the existing -enable-unsafe-fp-math option?

This is being driven by numerics people (picky as always),
believe it or not, and -enable-unsafe-fp-math does other things that  
they don't want.




More information about the llvm-commits mailing list