ci: Add automated benchmarking workflow #1365
Conversation
|
Compiler PytketIBMQ is best, or equal best on, 14.912280701754383% for circuits. |
|
Compiler PytketIBMQ is best, or equal best on, 14.912280701754383% for circuits. |
| - name: Checkout tket | ||
| uses: actions/checkout@v4 | ||
| with: | ||
| path: tket |
There was a problem hiding this comment.
Why is this necessary, as we are just using the downloaded wheel?
There was a problem hiding this comment.
Ah yes, a bit of a left over from testing. Cheers!
| # Ensure wheels are compatible with MacOS 12.0 and later: | ||
| export WHEEL_PLAT_NAME=macosx_12_0_arm64 |
There was a problem hiding this comment.
This shouldn't be necessary as we're testing on macos-14.
There was a problem hiding this comment.
Would you mind saying a little bit more about that. I'm afraid I'm not so familiar with the steps required to build the wheels.
I tried removing these two lines but see that the build then fails. I'm not sure if I should have expected that?
There was a problem hiding this comment.
Hmm, that's a surprise. Ah well, we can keep the lines there, it's not an issue.
daniel-mills-cqc
force-pushed
the
automatic_benchmarks
branch
from
ecca33b to
938082a
Compare
|
Compiler PytketIBMQ is best, or equal best on, 14.912280701754383% for circuits. |
Description
Adds a workflow to run benchmarks on each new PR. The version of pytket built from the PR is used in the benchmarks. At the moment this is a simple comparison to Qiskit on quantum volume circuits.
I would like to have a wider conversation about the compilers to compare agains, and the circuits to compile. The outcome of those conversation would feature in later PRs.
I also note that the benchmarks are quite slow. I will experiment with parallelisation in future PRs.
Related issues
N/A
Checklist