multi buffer: completed job return guarantees #174

Open
opened 2026-02-26 12:50:36 -06:00 by pobrn · 2 comments
pobrn commented 2026-02-26 12:50:36 -06:00 (Migrated from github.com)

Given a multi buffer manager with a runtime parallelism of n (i.e. it can work on n buffers in parallel), is it guaranteed that submitting n jobs will return at least one? Or does one need to submit at least n+1 jobs in the worst case (empty manager)?

It seems to me that submitting n jobs is enough to get a completed job back even if the manager is empty; but I would like to get confirmation that this is indeed guaranteed.

Given a multi buffer manager with a runtime parallelism of `n` (i.e. it can work on `n` buffers in parallel), is it guaranteed that submitting `n` jobs will return at least one? Or does one need to submit at least `n+1` jobs in the worst case (empty manager)? It seems to me that submitting `n` jobs is enough to get a completed job back even if the manager is empty; but I would like to get confirmation that this is indeed guaranteed.
pablodelara commented 2026-02-27 03:09:32 -06:00 (Migrated from github.com)

Hi @pobrn. Correct, it is guaranteed that after submitting those n jobs, you would get a completed job back, as long as you have submitted all the entire buffers (with ISAL_HASH_ENTIRE).

Hi @pobrn. Correct, it is guaranteed that after submitting those `n` jobs, you would get a completed job back, as long as you have submitted all the entire buffers (with ISAL_HASH_ENTIRE).
pobrn commented 2026-02-27 11:43:21 -06:00 (Migrated from github.com)

Does this not apply regardless of the submission flags? E.g. if n count of ISAL_HASH_FIRST jobs are submitted, then it is technically not guaranteed that at least one will be returned?

I suppose my question really is: what is the minimum number of jobs that one needs to have so that there is always at least one that the manager does not hold on to, which can be resubmitted with further data?

Does this not apply regardless of the submission flags? E.g. if `n` count of `ISAL_HASH_FIRST` jobs are submitted, then it is technically not guaranteed that at least one will be returned? I suppose my question really is: what is the minimum number of jobs that one needs to have so that there is always at least one that the manager does not hold on to, which can be resubmitted with further data?
Sign in to join this conversation.
No milestone
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
gh-intel/isa-l_crypto#174
No description provided.