-
Bug
-
Resolution: Duplicate
-
P4
-
None
-
8
-
None
-
generic
-
generic
Was NASHORN-842.
The mandreel test has a lot of global state that the current test harness (run-octane.js) can't successfully dismantle. Thus it can only do one run, at the second round it spins forever. This is not a bug, but the test is written that way. We should investigate if we can modify it a similar was as e.g. box2d (replace the teardown method, see run-octane.js) to get it to do multiple runs.
The octane tests were originally meant for single runs, but in order to really see what adaptive optimization can do to performance, we need to run them multiple times.
The mandreel test has a lot of global state that the current test harness (run-octane.js) can't successfully dismantle. Thus it can only do one run, at the second round it spins forever. This is not a bug, but the test is written that way. We should investigate if we can modify it a similar was as e.g. box2d (replace the teardown method, see run-octane.js) to get it to do multiple runs.
The octane tests were originally meant for single runs, but in order to really see what adaptive optimization can do to performance, we need to run them multiple times.
- duplicates
-
JDK-8014219 base.js in Octane benchmarks has way too much deviation
- Resolved