Worker API should provide safe defaults that increase performance in most cases

[runningcode]

When enabling worker APIs in third party plugins, such as kapt with kapt.use.worker.api=true, We have noticed slower build speeds due to workers consuming extra memory. Our builds seen an increase in time spent garbage collecting due to the extra workers and end up being slower.

It would be nice if Gradle would have safe defaults for the number of workers that would increase performance in most scenarios. Many teams do not have dedicated build engineers or resources to dedicate to this so it would be nice if the default would improve performance in most projects. Teams with additional resources would then be free to further optimize these values.

Let me know what additional information I can/should provide here.

My understanding is that the current default is # of CPUs. Perhaps a safer default would be # of CPUs / 2? Just tossing ideas out there.

[gavra0]

This could be handled automatically by Gradle, or plugin authors can provide CPU/memory hints for the worker actions. Using those resource usage hints, scheduling can be more efficient.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. Given the limited bandwidth of the team, it will be automatically closed if no further activity occurs. If you're interested in how we try to keep the backlog in a healthy state, please read our blog post on how we refine our backlog. If you feel this is something you could contribute, please have a look at our Contributor Guide. Thank you for your contribution.

[gildor]

It's still an important issue, now it requires multiple manual measurements and still not dynamic enough

[stale]

This issue has been automatically marked as stale because it has not had recent activity. Given the limited bandwidth of the team, it will be automatically closed if no further activity occurs. If you're interested in how we try to keep the backlog in a healthy state, please read our blog post on how we refine our backlog. If you feel this is something you could contribute, please have a look at our Contributor Guide. Thank you for your contribution.

[gildor]

It still looks like an important feature. It's hard now to measure such things and detect issues with memory and parallelism.