python -m eon.server --continuous generally scales quite nicely. e.g.
However, in practice, because of the way polling occurs at the start of each new phase, single laggards can lead to quite a lot of dead time.
To prevent this, we need bidirectional communication basically, so jobs can be submitted constantly as soon as more compute becomes free.
python -m eon.server --continuousgenerally scales quite nicely. e.g.However, in practice, because of the way polling occurs at the start of each new phase, single laggards can lead to quite a lot of dead time.
To prevent this, we need bidirectional communication basically, so jobs can be submitted constantly as soon as more compute becomes free.