Closed Jacfomg closed 11 months ago
Circuits are compiled to a list of sequences and then it is just using platform.execute_pulse_sequences
to deploy them. So it should use the same batch size as the sequences by default. Did you try an example and it didn’t work?
Yes, then it is my bad, the memory limit could be quite wrong when executing things with some several drive pulses. We may want to take them into account.
Yes, this is also something we need to check. For qblox I didn’t notice much difference from drive pulses but I didn’t try very very many.
I was running some RB with unrolling and with decent size depth for the circuits it reduced by a factor of 5 to 10 compared to simpler calibration sequences like ramsey.
Can we make circuit unrolling use the some kind of batches to fit in memory like we did for the sequences @stavros11 ?