search for: 060955

I am very much in favor of having a div instruction with well defined div-by-zero and overflow behavior. The undefined behavior on certain values for LLVM intrinsics has been a major pain point for us in Julia, because adding the extra branches just kills performance and we know that there is an X86 instruction that just does what we want. Anyway, this was brought up briefly above, but want to

...here. Thus, we'll keep invariant X = div(X,Y)*Y + rem(X,Y). > > > > > > > Does the patch look good? I'd be glad to hear any remarks and comments > regarding the patch, and adjust it correspondingly. > > [1] http://lists.cs.uiuc.edu/pipermail/llvmdev/2013-April/060955.html > > Best regards, > Michael > > On Apr 27, 2014, at 2:45 AM, Keno Fischer > <kfischer at college.harvard.edu <mailto:kfischer at college.harvard.edu>> > wrote: > >> I am very much in favor of having a div instruction with well defined >> div-b...