Skip to content

Add information about QPU contention #5336

Description

@jyu00

URL to the relevant guide or resource

Not sure, maybe https://quantum.cloud.ibm.com/docs/en/guides/monitor-job?

Select all that apply

  • new content request
  • typo
  • code bug
  • out-of-date content
  • broken link
  • other

Describe the fix or the content request.

Users often ask about why their job that should only take few seconds spend far longer in the "in progress" state. Some would even take the drastic step of canceling these jobs, fearing they'd be spending too much.

The docs on execution mode talk briefly about parallel classical processing, but it doesn't talk about QPU contention at all. We should add a section or a page explaining that:

  1. Each job needs to go through classical pre-processing before running on the QPU
  2. A job transitions to "in progress" status on IQP ("running" in Qiskit) as soon as its classical processing starts
  3. Most of the classical processing can be done in parallel, which is why multiple jobs can enter the "in progress" status simultaneously
  4. However, only one job can run on the QPU at a time. Hence QPU contention could happen if multiple jobs have gone through pre-processing and are ready for the QPU. If there is high QPU contention, a job can remain in "in-progress" state for longer than expected.
  5. Both the estimated usage and maximum execution time are based on the time the QPU is locked to execute the job and therefore exclude multi-threaded classical processing.

For new content requests - if the request is accepted, do you want to write the content?

I can help the team by providing enough information or by referring them to a specific Subject Matter Expert to write the material

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    Status
    In Progress

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions