Cryoris
commented
3 weeks ago I could reproduce this issue but the full Rust backtrace didn't make me any smarter:
``` run_in_parallel: true Starting layout length: 21 thread '
Running with RUST_BACKTRACE=1 seems to give a bit more sensible information, but I'm not sure how this is consistent with the full backtrace:
```
run_in_parallel: true
Starting layout length: 21
thread '' panicked at 'called `Option::unwrap()` on a `None` value', crates/accelerate/src/sabre/layer.rs:73:52
stack backtrace:
0: _rust_begin_unwind
1: core::panicking::panic_fmt
2: core::panicking::panic
3: qiskit_accelerate::sabre::route::swap_map_trial
4: qiskit_accelerate::sabre::layout::layout_trial
5: rayon::iter::plumbing::bridge_producer_consumer::helper
6: rayon_core::join::join_context::{{closure}}
7: rayon::iter::plumbing::bridge_producer_consumer::helper
8: rayon_core::join::join_context::{{closure}}
9: rayon::iter::plumbing::bridge_producer_consumer::helper
10: rayon_core::join::join_context::{{closure}}
11: rayon::iter::plumbing::bridge_producer_consumer::helper
12: rayon_core::join::join_context::{{closure}}
13: as rayon_core::job::Job>::execute
14: rayon_core::registry::WorkerThread::wait_until_cold
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
```
Environment
What is happening?
When running a large quantum volume circuit through the
SabreLayoutpass the individual trials start to internally panic with the following:once for each thread launched. Once all 20 threads panic the pass stops.
How can we reproduce the issue?
What should happen?
The pass shouldn't panic, and should successfully apply a layout and route the circuit.
Any suggestions?
This was most likely a regression introduced by: https://github.com/Qiskit/qiskit/pull/11977 as the example works fine with Qiskit 1.0.2.