Details
-
Sub-task
-
Resolution: Fixed
-
P4
-
23
-
b17
Description
Currently, we pattern match the addresses many times, sometimes in loops that are nested. We can create a VLoopAnalyzer module that caches the VPointers.
This might improve runtime, but it remains to be shown.
Creating such cached records might also allow us to sort them, and create the adjacent memory pair more efficiently in the future.
This might improve runtime, but it remains to be shown.
Creating such cached records might also allow us to sort them, and create the adjacent memory pair more efficiently in the future.
Attachments
Issue Links
- relates to
-
JDK-8330991 C2 SuperWord: refactor VPointer
- Open