Open levrik opened 1 year ago
It looks like the benchmark simply times out, pushing only a single number to times
list in run_benchmark
method which isn't supported when being passed to stdev
further down in the method. I guess that too many iterations are being executed per round. I'll be trying to dig deeper tomorrow.
Makes sense - I've never run a benchmark that takes the entire allowed time.
PR welcome to fix this.
I saw that pytest-benchmark
only runs a single iteration per round for "slow" running tests.
WDYT about implementing this? I would be open to prepare a PR.
That's basically the same as fixing the above error I guess.
Happy to review a PR.
I'm trying to collect more details but maybe you have an idea already. When running on a benchmark which according to
pytest_benchmark
takes ~0.32 ms per iteration, it fails with the following error:Another benchmark which is faster per iteration doesn't raise this error at all.