-
Enhancement
-
Resolution: Not an Issue
-
P3
-
repo-panama
The var-handle based memory access API is currently slower than a plain Unsafe access. This seems to be caused by two factors:
* the JIT has trouble in hoisting Scope liveness checks outside a loop
* when addresses are non-constant (e.g. because memory is allocated and accessed inside a try-with-resources), it seems like bound checks are also not optimized.
* the JIT has trouble in hoisting Scope liveness checks outside a loop
* when addresses are non-constant (e.g. because memory is allocated and accessed inside a try-with-resources), it seems like bound checks are also not optimized.
- relates to
-
-
- Resolved
-
-
-
- Resolved
-
-
JDK-8132243 Optimize Final Field Loads In Generated Code
-
- Draft
-