Timing the different proving steps

[mmagician]

If I'm not mistaken, the benchmark results give times for:

• Generate batch proof 0
• Generate batch proof 1 + Generate batch proof 01
• Generate batch proof 2
• Generate batch proof 3
• Generate batch proof 23
• Generate batch proof 03 + final proof

Is this the desired behavior? I would instead expect the following:

• Generate batch proof 0
• Generate batch proof 1
• Generate batch proof 01
• Generate batch proof 2
• Generate batch proof 3
• Generate batch proof 23
• Generate batch proof 03
• Generate final proof

It might be worth adding headers to the .csv.

[mmagician]

Ah, and also would be great to add the measurement units (ms). Without looking at the patched code it wasn't clear at all, especially since the whole process took a long time.

[StefanosChaliasos]

Feel free to add a header including ms.

I will need some time to double-check what times we keep track of. The point is that for our benchmarks, we only care about the first proof, one aggregation, and the final (compression) proof. This is because we only set specific inputs for the first one.