Reproduce 0.1 snapshot

[alecandido]

With the last modifications in #1088, I'm pretty sure that the sequence uploaded is fully equivalent to the 0.1 (exact same waveforms' samples, the only differences in instructions are loops of length 1, or set_ph 0 right after a reset_ph).

So, I'm now attempting to reproduce the exact same cluster snapshot (obtainable with cluster.print_readable_snapshot(), to avoid parsing manually all the Qcodes structures).

Here are the current snapshots being compared, before this PR.
0.1-snapshot.txt
0.2-snapshot.txt

The plan is to include all the differences, to make the 0.2 snapshot identical. The patches will be applied:

• directly in Qblox #1088, for those acknowledged to be required (i.e. whose role is explicitly understood) - and I will keep rebasing this branch
• or in here, if it is not clear how/why the change is affecting

Known differences

Cluster level

• reference_source, reference clock

Module level

• outX_offset, to replace what is currently Sequencer.offset_awg_path0
  • I know the second can be affected in real time, but I'm not sure how the value set via Qcodes acts differently from the module level one
• apply default values for _path, _offset, and scope_

Unused modules

They are not being configured in any way currently.

Sequencer level

• connect_outX, sometimes unset instead of I/Q/IQ/off
• marker_ovr_en, sometimes None instead of True
• marker_ovr_value, sometimes 0 instead of 15
• mod_en_awg, sometimes False instead of True
• nco_freq, sometimes None instead of 0
• sync_en, sometimes None instead of True
• thresholded_acq_threshold for the used qrm_rf sequencer disagrees (0.0060097 instead of 2.8846)
• thresholded_acq_rotation and thresholded_acq_threshold set for unused module (270.14 (Degrees) and -7.3504e-05)

Unused sequencers

They are not being configured in any way currently.

• default for all sequencers (cont_mode_..., mixer_..., ...) not applied

Updated snapshot

0.2-updated.txt

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 86 lines in your changes missing coverage. Please review.

Project coverage is 45.43%. Comparing base (0864f9a) to head (db8b79d).
Report is 1 commits behind head on qblox.

Files with missing lines	Patch %	Lines
src/qibolab/_core/instruments/qblox/config.py	0.00%	77 Missing ⚠️
src/qibolab/_core/instruments/qblox/cluster.py	0.00%	9 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##            qblox    #1159      +/-   ##
==========================================
- Coverage   46.21%   45.43%   -0.78%     
==========================================
  Files          96       96              
  Lines        4490     4567      +77     
==========================================
  Hits         2075     2075              
- Misses       2415     2492      +77     

Flag	Coverage Δ
unittests	45.43% <0.00%> (-0.78%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[alecandido]

Until 16aeec2, no difference with #1088

[alecandido]

@stavros11 I obtained virtually identical snapshots

Residual differences

The remaining differences are due to two crossed sequencers, since I'm using them incrementally (0, 1, 2, ...), while the 0.1 driver was using a default port assignment. However, these crossed sequencers are only seen for flux channels, which are not even involved in single shot.
The port assignment is also supposed to be crossed consistently, by the following invocation:

qibolab/src/qibolab/_core/instruments/qblox/config.py

Line 233 in 0102e71

seq.connect_sequencer(address.local_address)

The other differences are the mod 12 seq 1, which is missing some default properties in 0.1. However, this sequencer is supposed to be inactive, and I did not bother implementing the logic "set these configs on all default QCM-RF sequencers (i.e. seq < 2) but not on those on module where there are other default sequencers, if the current one is inactive".

The remaining present: False may be dropped, but they are not set explicitly by the 0.2 driver, but just indirectly by checking the module presence in the default round of settings. However, the difference is a None for a False in the snapshot for absent modules, which I'm pretty sure should not affect anything at all.

The image is obtained by:

• printing a cluster.print_readable_snapshot() just before beginning the execution
• manually replacing the name of the cluster in 0.2 (to make it equal to the 0.1 one, and reducing the subsequent diff)
• using difft, to obtain a more readable result

However, the reports are still significantly different.

Reports

You can notice some resemblance in the two pictures, and it seems clear that in the 0.2 one the ground state is being measured twice.

0.1	0.2

So, following the diff and the reports, it seems that there is no discrepancy in the configurations, and the acquisition is also good enough.

At this point, it seems that the drive is not being properly sent, and it's likely to be on the sequence side, rather than a configuration problem.

[alecandido]

And here are the comparisons for the drivers' sequences

Programs

0.1

0.2

# setup move 0, R1 # navgs loop nop loop_R1: move 0, R0 # nshots loop nop loop_R0: wait_sync 4 reset_ph set_ph 0 # set # relative phase 0 rads play 0,1,40 # play # waveforms DrivePulse(0, 40, # 0.355315, 4_103_156_133, # 0, Gaussian(5), L4-15, 0) # wait 200700 ns move 3, R3 wait_loop_R3: wait 65532 loop R3, @wait_loop_R3 wait 4104 add R0, 1, R0 nop jlt R0, 5000, @loop_R0 # nshots loop add R1, 1, R1 nop jlt R1, 1, @loop_R1 # navgs loop # cleanup stop *partially doctored, removing some whitespaces and wrapping some comments (just for the sake of readability side-by-side)

move 0,R0 # init bin counter move 0,R1 # init bin reset move 5000,R2 # init shots counter wait_sync 4 start: wait 4 play 0,1,40 wait 3395 # relaxation move 3,R3 wait3: wait 65535 loop R3,@wait3 reset_ph # phase reset add R0,1,R0 # bin increment loop R2,@start # loop over shots stop

Sequencer snapshots

0.1

0.2

qblox_controller_module12_sequencer0: parameter value ------------------------------------------------------------ connect_out0 : IQ connect_out1 : off cont_mode_en_awg_path0 : False cont_mode_en_awg_path1 : False cont_mode_waveform_idx_awg_path0 : 0 cont_mode_waveform_idx_awg_path1 : 0 gain_awg_path0 : None gain_awg_path1 : None marker_ovr_en : True marker_ovr_value : 15 mixer_corr_gain_ratio : 1 mixer_corr_phase_offset_degree : 0 mod_en_awg : True nco_freq : -202252867 (Hz) nco_phase_offs : 0 (Degrees) nco_prop_delay_comp : None (ns) nco_prop_delay_comp_en : None (ns) offset_awg_path0 : 0 offset_awg_path1 : 0 sync_en : True trigger10_count_threshold : None trigger10_threshold_invert : None trigger11_count_threshold : None trigger11_threshold_invert : None trigger12_count_threshold : None trigger12_threshold_invert : None trigger13_count_threshold : None trigger13_threshold_invert : None trigger14_count_threshold : None trigger14_threshold_invert : None trigger15_count_threshold : None trigger15_threshold_invert : None trigger1_count_threshold : None trigger1_threshold_invert : None trigger2_count_threshold : None trigger2_threshold_invert : None trigger3_count_threshold : None trigger3_threshold_invert : None trigger4_count_threshold : None trigger4_threshold_invert : None trigger5_count_threshold : None trigger5_threshold_invert : None trigger6_count_threshold : None trigger6_threshold_invert : None trigger7_count_threshold : None trigger7_threshold_invert : None trigger8_count_threshold : None trigger8_threshold_invert : None trigger9_count_threshold : None trigger9_threshold_invert : None upsample_rate_awg_path0 : 0 upsample_rate_awg_path1 : 0

iqm5q_qblox_module12_sequencer0: value ----------------------------- IQ off False False 0 0 None None True 15 1 0 True -202252867 (Hz) 0 (Degrees) None (ns) None (ns) 0 0 True None None None None None None None None None None None None None None None None None None None None None None None None None None None None None None 0 0

Waveforms

0.1

{
    'Envelope_Waveform_I(num_samples = 40, amplitude = 0.355315, shape = Gaussian(5))': {
        'data': [
            0.018216030410612587,
            0.024512348186780185,
            0.032473585278501534,
            0.04235353963840072,
            0.054383021382398256,
            0.06874658057406086,
            0.08555650186458075,
            0.10482601734711296,
            0.12644431305041612,
            0.15015633703066492,
            0.17555054343179538,
            0.20205743692669842,
            0.22896107208413613,
            0.255424533524822,
            0.28052896964037005,
            0.3033241386392847,
            0.3228868648375512,
            0.3383825290191143,
            0.34912394352834497,
            0.3546218456516639,
            0.3546218456516639,
            0.34912394352834497,
            0.3383825290191143,
            0.3228868648375512,
            0.3033241386392847,
            0.28052896964037005,
            0.255424533524822,
            0.22896107208413613,
            0.20205743692669842,
            0.17555054343179538,
            0.15015633703066492,
            0.12644431305041612,
            0.10482601734711296,
            0.08555650186458075,
            0.06874658057406086,
            0.054383021382398256,
            0.04235353963840072,
            0.032473585278501534,
            0.024512348186780185,
            0.018216030410612587
        ],
        'index': 0
    },
    'Envelope_Waveform_Q(num_samples = 40, amplitude = 0.355315, shape = Gaussian(5))': {
        'data': [
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0,
            0.0
        ],
        'index': 1
    }
}

0.2

Q1Sequence(
    waveforms={
        (UUID('faed3019-184e-4fe6-bde9-63de77e2d92f'), 0): Waveform(
            data=array([0.01821603, 0.02451235, 0.03247359, 0.04235354, 0.05438302,
       0.06874658, 0.0855565 , 0.10482602, 0.12644431, 0.15015634,
       0.17555054, 0.20205744, 0.22896107, 0.25542453, 0.28052897,
       0.30332414, 0.32288686, 0.33838253, 0.34912394, 0.35462185,
       0.35462185, 0.34912394, 0.33838253, 0.32288686, 0.30332414,
       0.28052897, 0.25542453, 0.22896107, 0.20205744, 0.17555054,
       0.15015634, 0.12644431, 0.10482602, 0.0855565 , 0.06874658,
       0.05438302, 0.04235354, 0.03247359, 0.02451235, 0.01821603]),
            index=0
        ),
        (UUID('faed3019-184e-4fe6-bde9-63de77e2d92f'), 1): Waveform(
            data=array([0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
       0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.,
       0., 0., 0., 0., 0., 0.]),
            index=1
        )
    },
    weights={},
    acquisitions={},
    program=Program(
        elements=[
            Line(Move(source=0, destination=Register(number=0), instr='move'), comment='init bin counter'),
            Line(Move(source=0, destination=Register(number=1), instr='move'), comment='init bin reset'),
            Line(Move(source=5000, destination=Register(number=2), instr='move'), comment='init shots counter'),
            Line(WaitSync(duration=4, instr='wait_sync')),
            Line(Wait(duration=4, instr='wait'), label='start'),
            Line(Play(wave_0=0, wave_1=1, duration=40, instr='play')),
            Line(Wait(duration=3395, instr='wait'), comment='relaxation'),
            Line(Move(source=3, destination=Register(number=3), instr='move')),
            Line(Wait(duration=65535, instr='wait'), label='wait3'),
            Line(Loop(a=Register(number=3), address=Reference(label='wait3'), instr='loop')),
            Line(ResetPh(instr='reset_ph'), comment='phase reset'),
            Line(Add(a=Register(number=0), b=1, destination=Register(number=0), instr='add'), comment='bin increment'),
            Line(Loop(a=Register(number=2), address=Reference(label='start'), instr='loop'), comment='loop over shots'),
            Line(Stop(instr='stop'))
        ]
    )
)

[alecandido]

Ok, the problem has been finally identified, and solved directly in #1088

It was due to the synchronization over different channels (a missing padding at the end of the experiment, since there is no align at the beginning).

Then, since it turned out to be completely unrelated to all the configurations, I will review the content in here to check if there is anything worth to be kept, and then close this PR.

[alecandido]

All the relevant changes have been ported to #1088, thus this PR is not any longer relevant.

(all the values not set explicitly, as it is done in here, are believed to assume those values by default internally, lacking a need to explicit enforce the defaults in this interface)