search for: pr16358

...iced in the same bit of code (PR17452). My suggestion was to be conservative in preserving NSW. It’s not about missing optimizations. I don’t have a concrete suggestion for harder problem of missing optimization. I can say that we need something better than embedding flags within SCEV expressions. PR16358 is a recent example of the general problem (there have been other unsolved cases that I can’t find now). Arnold made the point here that SCEV needs to infer the legal range of one expression from another. My very abstract suggestion for future improvements are: - At the SCEV level, logic to deduce...

On 10/03/2013 01:22 AM, Andrew Trick wrote: > > I’m not sure how to make sense of an NUW flag coming from a sub. > > NSW is just a misnomer for signed overflow. SCEV canonicalizes sub a,b to add a, (-b). Unfortunately, signed overflow is not the same thing for sub and add... > > sub nsw %a, %b != add nsw %a, (-1 * %b) > > sub nsw i32, -1, INT_MIN is true. > > add

...e same bit of code (PR17452). My suggestion was to be conservative in preserving NSW. It’s not about missing optimizations. > > I don’t have a concrete suggestion for harder problem of missing optimization. I can say that we need something better than embedding flags within SCEV expressions. PR16358 is a recent example of the general problem (there have been other unsolved cases that I can’t find now). Arnold made the point here that SCEV needs to infer the legal range of one expression from another. > > My very abstract suggestion for future improvements are: > - At the SCEV level,...