You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Some actors claim large resources for large swathes of time, but sit idle for lots of time as they wait for the deps to finish. This means we lose resource utilisation. Since the application of tasks is largely linear, we could refactor into a single actor that calls tasks. See visual below:
Currently if we have actors A (resources=X), B (resources=Y), C (resources=Z), linearly composed, A o B o C, then we claim X+Y+Z resources, even though only one is ever using compute at a given moment.
Instead we could reformulate as stitching together ephemeral tasks, freeing up resources in between.
Or, instead we could
The text was updated successfully, but these errors were encountered:
Some actors claim large resources for large swathes of time, but sit idle for lots of time as they wait for the deps to finish. This means we lose resource utilisation. Since the application of tasks is largely linear, we could refactor into a single actor that calls tasks. See visual below:
Currently if we have actors
A (resources=X), B (resources=Y), C (resources=Z)
, linearly composed,A o B o C
, then we claimX+Y+Z
resources, even though only one is ever using compute at a given moment.Instead we could reformulate as stitching together ephemeral tasks, freeing up resources in between.
Or, instead we could
The text was updated successfully, but these errors were encountered: