I have the following audit benchmark result in one of my PC, please if someone from team explain us how to interpret the result.
I have 3 questions to start with please.
- What is difference between single and rayon? Why the benchmark run both?
- What is the threshold to conclude the prove is too slow?
- Can we calculate the max plot size based on the result below? How?
PS H:\> .\subspace-farmer3g_9Nov benchmark prove D:\1
Benchmarking prove/plot/single: Warming up for 3.0000 s
Warning: Unable to complete 10 samples in 5.0s. You may wish to increase target time to 40.9s.
Benchmarking prove/plot/single: Collecting 10 samples in estimated 40.889 s (10 iterations)
Benchmarking prove/plot/single: Analyzing
prove/plot/single time: [3.2969 s 3.6483 s 3.9463 s]
change: [-18.980% -9.4513% -2.0186%] (p = 0.03 < 0.05)
Performance has improved.
Benchmarking prove/plot/rayon: Warming up for 3.0000 s
Warning: Unable to complete 10 samples in 5.0s. You may wish to increase target time to 5.2s.
Benchmarking prove/plot/rayon: Collecting 10 samples in estimated 5.1903 s (10 iterations)
Benchmarking prove/plot/rayon: Analyzing
prove/plot/rayon time: [485.30 ms 519.64 ms 552.07 ms]
change: [-18.094% -11.037% -4.7907%] (p = 0.00 < 0.05)
Performance has improved.
Same answer as Audit benchmark result - explanation please - #2 by nazar-pc
As described in How much time is given to submit a proof in 3g? - #2 by nazar-pc you have ~4 seconds to audit and prove, so depending on how fast auditing is, the remaining time will be used for proving. In practice since you need to audit within 1 second, you have 3s+ to prove.
No, proving performance doesn’t depend on plot size. The only exception here is when you have more than one eligible solution per slot, but in production network this is quite unlikely to happen, so you can more or less ignore it.
Thank you so much. Yes, huge difference between single and rayon.
Thanks God you and team figured out the root cause of Windows. No wonder last time most of us cannot farm effectively with large plot (4TB, 8TB). Yes, I understand proofing time is not impacted by plot size. But if plot is large, the time for audit can be more than 1s (e.g. up to 2s or 2.5s), then no time left for proofing.
If auditing takes more than one second, it will still farm, farmer would just have to skip some slots that is is not able to audit in time and yes, there will be less time for proving left in that case too. So it is more rational in that case to actually have twice as small plot size rather than twice as large and skip half of the slots + having difficulties proving them anyway due to reduce time window.