Open mikeharder opened 2 years ago
I think we will need to come up with a way to convert the CPU to some non-percentage value before publishing (e.g. x.y cores used). Otherwise if we ever change the VM SKU the metrics will not be historically comparable.
The perf automation should collect CPU and memory metrics of the spawned process(es), and include it in the results, like so:
For JS and Python, we will need to sum the memory and CPU of the child processes.