Solver failed with error: Not enough quota for metric 'fpga_job_hours' to process job.

Question

I have been evaluating the Quantum Inspired Optimizers available in Azure Quantum. I tested the Simulated Annealing solver first using the CPU version and then the FPGA hardware. Unfortunately I consumed all of my allocated FPGA time. Is it possible to get an extension to this limit?

However, my main problem is that now I get an error message when I switch back to executing on CPU hardware.

solver = SimulatedAnnealing(self.workspace, timeout=timeout, seed=22, platform=HardwarePlatform.CPU)

gives:
Solver failed with error: Cannot retrieve results as job execution failed (status: Failed. error: {'additional_properties': {}, 'code': 'NotEnoughQuota', 'message': "Not enough quota for metric 'fpga_job_hours' to process job. Current amount remaining is '0' but '-1' is required to proceed. ('-1' indicates that the amount required was not computed). Note that amount units depend on the dimension. Please contact 'Microsoft' for more information."})

even though I have not selected to execute on FPGA hardware. I get the same error even if I use the ParallelTempering solver.

Is there a way to continue to use CPU solvers when FPGA hardware quota has been exceeded?
And can I get my FPGA quota extended to I can continue to evaluate these solvers?

Thanks for your help.

Answer 1

Thank you for reaching out!

Is there a way to continue to use CPU solvers when my FPGA hardware quota has been exceeded?
Hitting the quota limit on FPGA should not block usage of CPU solvers. I'm going to follow up to get some information from you so that we can investigate this further.

Edit: Microsoft update - "The issue has been identified and we are working to roll-out a fix in the coming week. Thanks for reporting the issue!"

Can I get my FPGA quota extended so I can continue to evaluate these solvers?
You can select "Scale" from Microsoft QIO which will include the same 1 free hour but extend the quota to 50,000 hours. "Scale" remains as pay as you go with no minimum commitment.

The pricing sheet has up to date pricing:
https://aka.ms/aq/pricing