Remove provider-specific fake backends,`FakeProvider` class and related tools in 1.0 #11376

ElePT · 2023-12-06T15:40:22Z

Summary

(2/2) --> This is the last PR of the second step of the FakeBackends refactoring epic, it removes the snapshot-dependent fake backends as well as the fake provider. The corresponding deprecation PR will follow on the 0.46 branch -> #11381.

Details and comments

On hold until:

are merged.

qiskit-bot · 2024-01-16T09:24:43Z

One or more of the the following people are requested to review this:

@Qiskit/terra-core

The IBM-specific fake backends in Qiskit are being removed in Qiskit 1.0 in Qiskit/qiskit#11376. The same fake backends have been moved to qiskit-ibm-runtime in Qiskit/qiskit-ibm-runtime#1140 and follow up PRs. Here the `qiskit` imports are replaced with `qiskit-ibm-runtime` imports in order to provide a smooth transition between Qiskit 0.45 and Qiskit 1.0. In the future, it may be preferable to use generic fake backends maintained in the repo and not depend on `qiskit-ibm-runtime` for them.

The IBM-specific fake backends in Qiskit are being removed in Qiskit 1.0 in Qiskit/qiskit#11376. The same fake backends have been moved to qiskit-ibm-runtime in Qiskit/qiskit-ibm-runtime#1140 and follow up PRs. Here the `qiskit` imports are replaced with `qiskit-ibm-runtime` imports in order to provide a smooth transition between Qiskit 0.45 and Qiskit 1.0. In the future, it may be preferable to use generic fake backends maintained in the repo and not depend on `qiskit-ibm-runtime` for them. Additionally, this commit reorders the requirements-dev.txt file so it is easier to read and adds qiskit-ibm-runtime as a development dependency.

ElePT · 2024-02-01T10:48:38Z

All the blocking PRs have been merged 🎉 . And the conflicts have been fixed, so removing the on hold label.

coveralls · 2024-02-01T11:20:57Z

Pull Request Test Coverage Report for Build 7749096260

0 of 1 (100.0%) changed or added relevant line in 1 file are covered.
138 unchanged lines in 6 files lost coverage.
Overall coverage decreased (-0.2%) to 89.164%

Files with Coverage Reduction	New Missed Lines	%
crates/qasm2/src/expr.rs	1	93.81%
qiskit/providers/models/backendproperties.py	1	94.82%
qiskit/providers/models/pulsedefaults.py	4	96.46%
crates/qasm2/src/lex.rs	7	90.93%
crates/qasm2/src/parse.rs	18	96.23%
qiskit/providers/fake_provider/fake_backend.py	107	39.13%

Totals
Change from base Build 7749016482:	-0.2%
Covered Lines:	58792
Relevant Lines:	65937

💛 - Coveralls

kevinhartman

Looks good to me! 😁

The underlying cause of the windows failures were actually much larger performance issues in aer's `NoiseModel.from_backend()` constructor method when run with a BackendV2 instance. This is causing the `NoiseModel` construction for `GenericBackendV2` to take a great deal of time and use up all the memory we have available in the windows VMs in CI. To avoid this issue in the short term until a fix can be added to aer this commit just limits the size of the backend v2 examples we run to prevent resources problems in CI.

The aer noise model construction was an artifact of using all to all connectivity in the test. Using a more sparse connectivity alleviates the performance issues. This commit switches to use a ring graph and restores the larger qubit count backends.

jakelishman

Ok, this is the one.

jakelishman · 2024-02-01T23:48:51Z

For posterity: we saw Azure went out of memory twice on Windows with Python 3.8 during attempts to merge this, which made us suspicious, but we thought it might just be Windows nonsense, so were going to skip it on Windows and sort it out later. The we found the test.python.providers.test_fake_backends file was taking nearly 10 minutes on my (not especially powerful) Windows machine, ~6 minutes on my Mac, and Matt's Linux box was taking a lot longer than normal too, so it was more than that.

We found that it was spending all the time in Aer building the noise model, which at first we thought was just inefficiencies in Aer (we used to build the noise model in house in the old fake backends) and were going to limit the tests to only 5q and 7q devices to sort it out on Aer.

Then we realised that it's spending all the time in the noise model because the GenericBackendV2(num_qubits=...) instances we were constructing to parametrise the tests over default to all-to-all coupling maps, which means that the noise-model construction is quadratic in the number of qubits (as well as Aer's calcs of it being not very efficient). The proper fix for us is to limit the connectivity of the devices under test (and probably in the future to improve Aer's handling of noise models).

The test in question was test.python.providers.test_fake_backends.TestFakeBackends.test_circuit_on_fake_backend_v2, though the problems really arose from the parametrisation of that test.

#1369) The IBM-specific fake backends in Qiskit are being removed in Qiskit 1.0 in Qiskit/qiskit#11376. The same fake backends have been moved to qiskit-ibm-runtime in Qiskit/qiskit-ibm-runtime#1140 and follow up PRs. Here the `qiskit` imports are replaced with `qiskit-ibm-runtime` imports in order to provide a smooth transition between Qiskit 0.45 and Qiskit 1.0. In the future, it may be preferable to use generic fake backends maintained in the repo and not depend on `qiskit-ibm-runtime` for them. Additionally, this commit reorders the requirements-dev.txt file so it is easier to read and adds qiskit-ibm-runtime as a development dependency. It also temporarily pins qiskit to 0.45 while other deprecations are addressed.

This change adjusts for several remaining changes in Qiskit 1.0: * qiskit.providers.basicaer was removed in Qiskit 1.0 (see Qiskit/qiskit#11422). Here its use is replaced with qiskit-aer. * `qiskit.providers.fake_provider.FakeJob` was removed in Qiskit 1.0 (see Qiskit/qiskit#11376). Here the `FakeJob` already in `qiskit_experiments.test` is used instead. This update was missed in f22bb7b which switched the other `fake_provider` usage. * Replace `qiskit.provider.FakeProvider` with a custom subclass of `ProviderV1`. `FakeProvider` was moved out of Qiskit 1.0 to qiskit-ibm-runtime. * `calibration/test_rabi.py` was modified to make the data for a test of bad data worse as the test started fitting the data as good with the original parameters and qiskit-aer in place of basicaer. * In the readout mitigation manual, a use of `plot_histogram` was exchanged for one of `plot_distribution`. * Guard `FakeBackendV2` import that was removed in Qiskit 1.0. Try to import `FakeBackendV2` from qiskit-ibm-runtime as well. Also, this change unpins qiskit which had to be pinned in f22bb7b until these fixes could be merged since 0.46.0 was released with deprecation warnings for some of the usage changed here.

ElePT added the on hold Can not fix yet label Dec 6, 2023

ElePT added this to the 1.0.0 milestone Dec 7, 2023

ElePT added the Changelog: Removal Include in the Removed section of the changelog label Dec 7, 2023

ElePT mentioned this pull request Dec 7, 2023

Add generic Fake Backends and update use in Qiskit unit tests #10954

Closed

7 tasks

ElePT changed the title ~~Remove provider-specific fake backends, fake provider and related tools in 1.0~~ Remove provider-specific fake backends,FakeProvider class and related tools in 1.0 Dec 7, 2023

ElePT mentioned this pull request Dec 7, 2023

Add generic V1 Fake Backends, replace use in tests #10952

Merged

9 tasks

ElePT marked this pull request as ready for review January 16, 2024 09:24

ElePT requested review from jyu00 and a team as code owners January 16, 2024 09:24

ElePT mentioned this pull request Jan 16, 2024

Deprecate provider-specific fake backends,FakeProvider class and related tools in 0.46 #11381

Merged

5 tasks

mtreinish assigned kevinhartman Jan 23, 2024

wshanks mentioned this pull request Jan 23, 2024

Switch tests and documentation to use qiskit-ibm-runtime fake backends qiskit-community/qiskit-experiments#1369

Merged

ElePT mentioned this pull request Jan 26, 2024

Add fake generic and modify tests #10266

Merged

Remove provider-specific fake backends, fake provider and related tools

5d8cb06

ElePT force-pushed the remove-fake-backends branch from 78cb3f3 to b6d5aa6 Compare February 1, 2024 10:46

ElePT requested a review from nonhermitian as a code owner February 1, 2024 10:46

ElePT removed the on hold Can not fix yet label Feb 1, 2024

Add release note

511e90c

ElePT force-pushed the remove-fake-backends branch from b6d5aa6 to 511e90c Compare February 1, 2024 10:51

Remove sneaky fake generic

f614a62

ElePT requested review from eggerdj and wshanks as code owners February 1, 2024 18:52

I'm sorry I'm at the point where I confuse 1s and 5s.

0c49870

kevinhartman previously approved these changes Feb 1, 2024

View reviewed changes

kevinhartman enabled auto-merge February 1, 2024 19:47

Fix skip message when aer isn't installed

3ca7445

mtreinish dismissed kevinhartman’s stale review via 3ca7445 February 1, 2024 20:52

mtreinish previously approved these changes Feb 1, 2024

View reviewed changes

Skip fake backend test with aer on windows

ac93ba1

mtreinish dismissed their stale review via ac93ba1 February 1, 2024 22:37

mtreinish previously approved these changes Feb 1, 2024

View reviewed changes

jakelishman disabled auto-merge February 1, 2024 23:21

mtreinish dismissed their stale review via f158512 February 1, 2024 23:25

mtreinish previously approved these changes Feb 1, 2024

View reviewed changes

mtreinish dismissed their stale review via 24343f9 February 1, 2024 23:37

mtreinish approved these changes Feb 1, 2024

View reviewed changes

Merge branch 'main' into remove-fake-backends

8163e32

jakelishman approved these changes Feb 1, 2024

View reviewed changes

mtreinish merged commit 84f0d9b into Qiskit:main Feb 2, 2024
12 checks passed

wshanks mentioned this pull request Feb 2, 2024

Remove remaining code usage deprecated in Qiskit 0.46 qiskit-community/qiskit-experiments#1384

Merged

ElePT mentioned this pull request Mar 12, 2024

Remove already "removed" FakeBackendV2 #11997

Merged

1ucian0 mentioned this pull request Apr 2, 2024

Creating a challenging backend to show the transpiler stochasticity #12116

Merged

mergify bot mentioned this pull request Apr 4, 2024

Creating a challenging backend to show the transpiler stochasticity (backport #12116) #12141

Merged

ohuettenhofer mentioned this pull request Oct 31, 2024

Fix current Qiskit versions tequilahub/tequila#368

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove provider-specific fake backends,`FakeProvider` class and related tools in 1.0 #11376

Remove provider-specific fake backends,`FakeProvider` class and related tools in 1.0 #11376

ElePT commented Dec 6, 2023 •

edited

Loading

qiskit-bot commented Jan 16, 2024

ElePT commented Feb 1, 2024

coveralls commented Feb 1, 2024 •

edited

Loading

kevinhartman left a comment

jakelishman left a comment

jakelishman commented Feb 1, 2024 •

edited

Loading

Remove provider-specific fake backends,FakeProvider class and related tools in 1.0 #11376

Remove provider-specific fake backends,FakeProvider class and related tools in 1.0 #11376

Conversation

ElePT commented Dec 6, 2023 • edited Loading

Summary

Details and comments

qiskit-bot commented Jan 16, 2024

ElePT commented Feb 1, 2024

coveralls commented Feb 1, 2024 • edited Loading

Pull Request Test Coverage Report for Build 7749096260

💛 - Coveralls

kevinhartman left a comment

Choose a reason for hiding this comment

jakelishman left a comment

Choose a reason for hiding this comment

jakelishman commented Feb 1, 2024 • edited Loading

Remove provider-specific fake backends,`FakeProvider` class and related tools in 1.0 #11376

Remove provider-specific fake backends,`FakeProvider` class and related tools in 1.0 #11376

ElePT commented Dec 6, 2023 •

edited

Loading

coveralls commented Feb 1, 2024 •

edited

Loading

jakelishman commented Feb 1, 2024 •

edited

Loading