Experiment i did is very simple (and honest i guess, without tricks): i've synchronized both daemons bitmonerod(prebuilt bitmonerod i've downloaded from website) and boolbd, and yesterday i've shutdowned them both in one time(the difference was maybe 1-2 minutes).
Interesting results!
One thing I noticed here:
1818 Blocks
...
750 Blocks
The first is Monero, the second is Boolberry.
Why does Monero have almost 2.5x as many blocks in the same time period (when normally one would expect 2x)? As far as I know they both use the same difficulty targeting algorithm so presumably this is due to variation in the network hash rates and realized block times as opposed to the target. This certainly doesn't invalidate your test results -- they were an accurate measurement for a particuar day -- but it calls into question whether this was a representative time interval. I would guess there will also be time periods when Monero has
fewer than the expected number of blocks being verified (do you agree?)
But in any case this significantly increases the number of PoW verifications to be done. You calculated this at 36.36 seconds, but normalizing for the number of blocks to 1500 (I think you would agree this is a reasonable adjustment?) this drops to 30 seconds. Ultimately that is a small difference though, because as I've mentioned on several different threads, the PoW time is a relatively small share of syncing time, and your results seem to confirm this (but see below).
Similarly, normalizing for number of blocks would reduce the amount of time spent "processing blocks" (managing the chain data, coinbase transactions, etc.) but would not affect the processing of (non-coinbase) transactions.
Unfortunately this means (as you said) it is hard to normalize further because we don't really know the breakdown between "processing blocks" and "processing transactions" and even if Monero had closer to the target number of blocks presumably it would have the same number of non-coinbase transactions.
One thing is still interesting to me though. You report 2.5 minutes of sync time for Monero for a 24 hour period, yet people often report many
hours to sync the chain for a month or less. There is a big gap there, that we are still working to understand it. This is what I meant by sync time is not being explained by the above factors. Adding approximately 2 minutes per day to sync time (somewhat less if we normalize for block count) does not explain hours and hours of syncing. Hopefully we can report more on this soon.
EDIT: Oh, I just thought of something else. If Monero has more 2.5x more blocks and the same number of tx then the number of tx per block should be 1/2.5x. Yet
excluding PoW time we see Monero taking
more time to process each block despite having
many fewer transactions in each block! This is quite unexpected and points to some other factor, probably the larger transaction history, but possibly some gross inefficiency in the scaling with respect to the size of transaction history. We should look at this more closely. After all, eventually both coins will have many more transactions in their history than they do now!