"Signed integers are two's complement" from CppCon is now online!
https://www.youtube.com/watch?v=JhUxIVf1qok …
-
-
We get back *a bit* of the benefits of overflow UB from GEP inbounds. e.g. the cited post about gcc references being able to know x[i] precedes x[i+1] and I'm pretty sure inbounds implies that (but idk how well that can back-prop to indexes themselves).https://kristerw.blogspot.com/2016/02/how-undefined-signed-overflow-enables.html …
-
Yeah, my guess would be that the vast majority of the cited 12% perf loss is the loop optimization problems, because those can have cascading effects on downstream optimizations. But I don’t really know.
- 5 more replies
New conversation -
-
-
Interestingly, it looks to me like Rust does miss out on some loop optimizations at the moment, comparing https://rust.godbolt.org/z/EpYTab and https://godbolt.org/z/eDMxZM (I'm guessing because SCEV can't determine trip count?), but that should be able to be improved in the future.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.