baloghadamsoftware added a comment. In https://reviews.llvm.org/D49074#1160793, @NoQ wrote:
> I'd also rather stick to integer arithmetic and avoid using floats even in > intermediate calculations. It'd be hard to make sure that no rounding errors > kick in if we use floats. Yes, I agree. I think an integer plus a flag (whether to take the reciprocal) is better since we only use n or 1/n for scaling. https://reviews.llvm.org/D49074 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
