Azure Quantum unveiled its newest feature called sessions, an added option that eliminates the requirement for users to possess knowledge about the cloud provider or quantum manufacturer. By making a Python API call, users can initiate a session that will categorize their tasks and utilize performance optimizations offered by the quantum manufacturer, making it both user-friendly and efficient.
In collaboration with Quantinuum, users will have the opportunity to gain exclusive access to the QPU for the duration of the session.
Sessions are equipped with a default “cancel-on-failure” policy for jobs to address this issue. If a position within a session fails, Azure Quantum will automatically cease submitting subsequent jobs and endeavour to cancel any jobs already submitted to a target. However, users can disable this policy if they wish to continue running subsequent jobs in their session despite previous job failures.
Enhanced Job Management and Processing with Sessions in Azure Quantum and Quantinuum
With sessions, it becomes possible to categorize jobs under a single run of an algorithm, making it simpler to manage and track them. Whenever jobs are managed within an Azure Quantum workspace through the Azure Portal, sessions are presented alongside any jobs submitted without a session. Accessing a session makes it possible to view the job specifics for all jobs in the session.
All jobs dispatched to Azure Quantum’s hardware partners come with session IDs that enable them to recognize associated jobs and optimize processing. In conjunction with introducing sessions, Quantinuum has added an impressive feature that can significantly enhance performance for researchers running complex hybrid algorithms.
Exclusive QPU Access for Complex Algorithms with Quantinuum
In the case of all Quantinuum targets, once the first job in a session is completed, any subsequent jobs submitted within a minute of the previous job execution are granted exclusive access to the target for the duration of the session.
Instead of queueing individual jobs, Quantinuum treats them as components of a single algorithm, and with the support of sessions, they receive dedicated QPU access.
Read more about it here.