ARM uses asymmetric core design https://developer.arm.com/documentation/den0024/a/Multi-core-processors/Multi-processing-systems/Asymmetric-multi-processing
Currently I see this feature available on macOS, but I'm guessing that the API to control which core to use will be eventually (already is?) available on all aarch64 platforms.
According to https://developer.apple.com/library/archive/documentation/Performance/Conceptual/power_efficiency_guidelines_osx/PrioritizeWorkAtTheTaskLevel.html
we can help the macOS run our VM more efficiently, by providing it with hints on how to prioritize aspects of our process, for example the priority of threads. We should investigate and use the provided facilities in macOS to set QoS for our VM threads as needed.
Currently I see this feature available on macOS, but I'm guessing that the API to control which core to use will be eventually (already is?) available on all aarch64 platforms.
According to https://developer.apple.com/library/archive/documentation/Performance/Conceptual/power_efficiency_guidelines_osx/PrioritizeWorkAtTheTaskLevel.html
we can help the macOS run our VM more efficiently, by providing it with hints on how to prioritize aspects of our process, for example the priority of threads. We should investigate and use the provided facilities in macOS to set QoS for our VM threads as needed.
- relates to
-
JDK-8273121 Investigate impact of big.little/hybrid archs on gc thread sizing
-
- Open
-