| Issue |
130639
|
| Summary |
Wider-than-native shifts are compiled inefficiently when the shift count is partially known
|
| Labels |
new issue
|
| Assignees |
|
| Reporter |
CatsAreFluffy
|
On 32-bit platforms, 64-bit shifts can be compiled more efficiently if you know whether or not the shiftcount is greater than 32 or not. For example, in [this Godbolt](https://godbolt.org/z/KM1bTsbjv), `short_shift` and `long_shift` should both be able to skip the `test cl, 32` instruction.
![]()
_______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
