Time to completition

We've just found and fixed a bug where a job that was stopped in the second half then restarted might show a percentage complete which is less than 50% for some time. This is because the restarted job resumes part of the 'first half' computation from the last saved checkpoint before going back to resuming the second half of the computation from a saved checkpoint. How large this effect is on any given machine depends upon how the user preferences have been set for checkpointing frequency / interval.

This will be fixed in our next app release.

Cheers,
Bruce