Zurich measurement sequences start at time set in Pulses #685
Conversation
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## main #685 +/- ##
==========================================
- Coverage 62.15% 62.09% -0.07%
==========================================
Files 49 49
Lines 6834 6857 +23
==========================================
+ Hits 4248 4258 +10
- Misses 2586 2599 +13
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
|
Also, this crashes when multiple measurements are applied on the same sequence. Let's say, readout_characterization routine for example. |
…m/qibolab into zh_measurement_sequences
Merge branch main
There was a problem hiding this comment.
Thanks @GabrielePalazzo the code looks good, and works as intented in the existing routines I tried. I just have some small format comments. I would test with an unrolling routine soon before approving.
| pulse_sequences.append([]) | ||
| measurement_index = 0 | ||
| for pulse in pulses: | ||
| if measurement_index < len(measurements): |
There was a problem hiding this comment.
| if measurement_index < len(measurements): | |
| if measurement_index < len(measurements) and isinstance(quantum_element, Qubit): |
How about something like this as we will never read couplers ?
Co-authored-by: Juan Cereijo <[email protected]>
|
I was trying to see if I could reproduce the same results with and without the unrolling using qiboteam/qibocal#629. As the qubits were not calibrated I could see anything so I did qiboteam/qibocal#665. The results seem the same so that means no phase reset on the measurament or whatever is needed. |
…b into zh_measurement_sequences
|
@GabrielePalazzo there is an issue with the T1 sweepers |
There was a problem hiding this comment.
Everything seems working as intended. The only thing left would be to update the number of sequences in a smart way they can take for the unrolling in a later benchmark. Thanks @GabrielePalazzo.
|
I merged main bcause #629 and found a typo on my code. |
There was a problem hiding this comment.
Thanks @GabrielePalazzo @Jacfomg. I tested this with allXY and unrolling and the results seem compatible with the no unrolling case, so I guess the measurement order has been fixed.
| @@ -55,8 +55,8 @@ | |||
| SWEEPER_BIAS = {"bias"} | |||
| SWEEPER_START = {"start"} | |||
|
|
|||
| MAX_MEASUREMENTS = 32 | |||
| """Maximum number of readout pulses in a single sequence.""" | |||
| MAX_SEQUENCES = 150 | |||
There was a problem hiding this comment.
I have not done extensive benchmarking about this, but as a general note it is better to underestimate than push it to its limit, because execution will fail completely if we exceed the memory. While if we underestimate, it may take a bit longer because it will split more batches but it will still work.
There was a problem hiding this comment.
Yeah I agree, I found that for RB like circuits it would be like 20 while for simple calibration experiments it was around 150. So it is a huge change, I would suggest to count the number of pulses in the sequence as part of the batching maybe.
Checklist: