Queue consolidation and peer-to-peer quantum workload stealing
Distributing quantum jobs are disclosed. When a quantum processing unit is underutilized or when wait times are long, quantum jobs may be distributed from the job queue of one vendor to another vendor. This improves utilization and reduces wait times.
1 . A method, comprising:
determining a utilization status of a first quantum processing unit when preparing to execute a quantum circuit included in a first quantum job retrieved from a first job queue associated with the first quantum processing unit;
evaluating a second job queue of a second quantum processing unit to determine if the job queue contains a second quantum job that can be executed concurrently with or separately from the first quantum job based on metadata describing a number of qubits required and entanglement characteristics of the second quantum job;
retrieving at least a portion of the second quantum job from the second job queue into the first job queue when unused qubits are available on the first quantum processing unit and the at least a portion of the second quantum job is compatible for concurrent execution based on qubit usage and when the qubits required by the first quantum job and the at least portion of the second quantum job are not entangled with each other; and
performing the first quantum job and at least the portion of the second quantum job in the first quantum processing unit, the first quantum job and the at least portion of the second quantum job being executed (i) in parallel when the qubit partitions of the first quantum job and the at least portion of the second quantum job are non-entangled and fit within unused qubits, or (ii) sequentially when parallel execution cannot be performed due to qubit availability or entanglement constraints.
2 . The method of claim 1 , further comprising evaluating a number of qubits required by the second quantum job and determining whether total qubits used by the first quantum job and the second quantum job are less than or equal to qubits offered by the first quantum processing unit.
3 . The method of claim 1 , further comprising distributing a second portion of the second quantum job to a third quantum processing unit, wherein the third quantum processing unit executes the second quantum job.
4 . The method of claim 3 , further comprising reducing user wait time by coordinating a split of the second quantum job into the first portion and the second portion.
5 . The method of claim 1 , further comprising reducing wait time associated with performing the second quantum job.
6 . The method of claim 1 , further comprising searching the second job queue for the second quantum job using a greedy search or a heuristic search.
7 . The method of claim 1 , further comprising evaluating the second job queue using a queue exchange.
8 . The method of claim 7 , wherein the queue exchange comprises the second quantum processing unit publishing descriptions of quantum jobs in the second job queue or the second quantum processing unit providing an API that allows the descriptions to be queried.
9 . The method of claim 1 , further comprising combining multiple circuits in the first and second quantum jobs to fit a qubit space of the first quantum processing unit.
10 . The method of claim 1 , further comprising selecting the second quantum job based on overall value to owners of the first and second quantum processing units and to owners of the first and second quantum processing jobs.
11 . A non-transitory storage medium having stored therein instructions that are executable by one or more hardware processors to perform operations comprising:
determining a utilization status of a first quantum processing unit when preparing to execute a quantum circuit included in a first quantum job retrieved from a first job queue associated with the first quantum processing unit;
evaluating a second job queue of a second quantum processing unit to determine if the job queue contains a second quantum job that can be executed concurrently with or separately from the first quantum job based on metadata describing a number of qubits required and entanglement characteristics of the second quantum job;
retrieving at least a portion of the second quantum job from the second job queue into the first job queue when unused qubits are available on the first quantum processing unit and the at least a portion of the second quantum job is compatible for concurrent execution based on qubit usage and when the qubits required by the first quantum job and the at least portion of the second quantum job are not entangled with each other; and
performing the first quantum job and the at least a portion of the second quantum job in the first quantum processing unit, the first quantum job and the at least portion of the second quantum job being executed (i) in parallel when the qubit partitions of the first quantum job and the at least portion of the second quantum job are non-entangled and fit within unused qubits, or (ii) sequentially when parallel execution cannot be performed due to qubit availability or entanglement constraints.
12 . The non-transitory storage medium of claim 11 , further comprising evaluating a number of qubits required by the second quantum job and determining whether total qubits used by the first quantum job and the second quantum job are less than or equal to qubits offered by the first quantum processing unit.
13 . The non-transitory storage medium of claim 11 , further comprising distributing a second portion of the second quantum job to a third quantum processing unit, wherein the third quantum processing unit executes the second quantum job.
14 . The non-transitory storage medium of claim 13 , further comprising reducing user wait time by coordinating a split of the second quantum job into the first portion and the second portion.
15 . The non-transitory storage medium of claim 11 , further comprising reducing wait time associated with performing the second quantum job.
16 . The non-transitory storage medium of claim 11 , further comprising searching the second job queue for the second quantum job using a greedy search or a heuristic search.
17 . The non-transitory storage medium of claim 11 , further comprising evaluating the second job queue using a queue exchange.
18 . The non-transitory storage medium of claim 17 , wherein the queue exchange comprises the second quantum processing unit publishing descriptions of quantum jobs in the second job queue or the second quantum processing unit providing an API that allows the descriptions to be queried.
19 . The non-transitory storage medium of claim 11 , further comprising combining multiple circuits in the first and second quantum jobs to fit a qubit space of the first quantum processing unit.
20 . The non-transitory storage medium of claim 11 , further comprising selecting the second quantum job based on overall value to owners of the first and second quantum processing units and to owners of the first and second quantum processing jobs.