[WIP] Add HLS plugin benchmarks

[ElePT]

Summary

This is a first attempt at adding a file to test potential performance improvements in HLS plugins.

Details and comments

[coveralls]

Pull Request Test Coverage Report for Build 8982169336

Warning: This coverage report may be inaccurate.

This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.

• For more information on this, see Tracking coverage changes with pull request builds.
• To avoid this issue with future PRs, see these Recommended CI Configurations.
• For a quick fix, rebase this PR at GitHub. Your next report should be accurate.

Details

• 0 of 0 changed or added relevant lines in 0 files are covered.
• 56 unchanged lines in 5 files lost coverage.
• Overall coverage increased (+0.02%) to 89.642%

---

Files with Coverage Reduction	New Missed Lines	%
qiskit/transpiler/passes/synthesis/unitary_synthesis.py	2	88.2%
crates/qasm2/src/lex.rs	4	92.37%
qiskit/converters/init.py	7	72.0%
crates/accelerate/src/sparse_pauli_op.rs	18	93.27%
crates/qasm3/src/circuit.rs	25	78.06%

Totals
Change from base Build 8966541600:	0.02%
Covered Lines:	62250
Relevant Lines:	69443

---

💛 - Coveralls

[mtreinish]

I would probably move this to setUp() to avoid this skewing the measurements. The entire contents of this function is what ends up being measured.

[mtreinish]

We don't need to do this explicitly the asv harness handles taking multiple samples to get the statistical sampling. We probably want to move the random clifford construction to setup too though to avoid randomization impacting the measurements.

[alexanderivrii]

Should we nonetheless consider multiple different random linear functions, random cliffords, etc. to avoid overtuning synthesis algorithms on specific circuits?

[ElePT]

Thanks for the suggestions. I have moved the circuit and pm construction to the setup and fixed the random seeds, and applied a few offline suggestions from @alexanderivrii. So far the benchmarks run but I wonder if the circuit I am currently building makes sense for the bm_wide_circuit test, because the run seems to be too fast (range of ms instead of s), so it's probably not doing any synthesis? The difference between ag/bm and default/greedy is also quite stark in the wide_circuit tests (maybe the first two are also not synthesizing). I will add tests that monitor depth/gate count to try to get more information:

[55.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_clifford_ag_rand_benchmarking           3.34±0.08s                                                                                                       
[60.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_clifford_ag_wide_circuit                1.24±0.08s                                                                                             
[65.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_clifford_bm_rand_benchmarking           9.98±0.09s                                                                                                       
[70.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_clifford_bm_wide_circuit                2.02±0.07ms                                                                                                   
[75.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_clifford_default_rand_benchmarking      9.47±0.2s                                                                                         
[80.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_clifford_default_wide_circuit           26.1±0.9s                                                                                                     
[85.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_clifford_greedy_rand_benchmarking       9.22±0.2s                                                                                                   
[90.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_clifford_greedy_wide_circuit            25.8±0.5s                                                                                               
[95.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_linear_func_kms                         1.86±0.1s                                                                                                   
[100.00%] ··· transpiler_hls_plugins.HLSPluginsSuite.time_linear_func_pmh                        1.14±0.06s